Test case that failed my balancer refactor https://github.com/coreos/etcd/pull/8834.
Current, kv network partition tests do not specifically test
isolated leader case.
This PR moves TestKVSwitchUnavailable to network_partition_test.go
and make it always isolate leader.
Signed-off-by: Gyu-Ho Lee <gyuhox@gmail.com>
Since 3-second is the minimum time to keep an endpoint in unhealthy,
it is possible that endpoint switch happens right after context timeout.
Signed-off-by: Gyu-Ho Lee <gyuhox@gmail.com>
RPC should be sent to trigger 'readyWait' on new pin address.
Otherwise, endpoints other than ep[0] may be pinned.
Signed-off-by: Gyu-Ho Lee <gyuhox@gmail.com>
'member' type is not exported.
In network partition tests, we want do
InjectPartition(t, clus.Members[lead], clus.Members[lead+1])
Signed-off-by: Gyu-Ho Lee <gyuhox@gmail.com>
When no address is pined, and balancer ignores the addr Up due to
its current unhealthy state, balancer will be unresponsive forever.
This PR fixes it by doing a full reset when there is no pined addr,
thus re-trigger the Up call.
When creating multiple clients, 'mustClientFromCmd' overwrites
inherited flags with environment variables, so later clients
were printing warnings on duplicate key updates.
Signed-off-by: Gyu-Ho Lee <gyuhox@gmail.com>
Previous behavior is when server returns errors, retry
wrapper does not do anything, while passively expecting
balancer to gray-list the isolated endpoint. This is
problematic when multiple endpoints are passed, and
network partition happens.
This patch adds 'endpointError' method to 'balancer' interface
to actively(possibly even before health-check API gets called)
handle RPC errors and gray-list endpoints for the time being,
thus speeding up the endpoint switch.
This is safe in a single-endpoint case, because balancer will
retry no matter what in such case.
Signed-off-by: Gyu-Ho Lee <gyuhox@gmail.com>
Otherwise network-partitioned member with active health-check
server would not be gray-listed, making health-balancer stuck
with isolated endpoint.
Also clarifies some log messages.
Signed-off-by: Gyu-Ho Lee <gyuhox@gmail.com>
Several goroutines may call setPrevRev concurrently with different
revisions, all higher than prevRev. Previously all of these goroutines
could set prevRev, so prevRev may be replaced by older one.
If response's revision equals to prevRev, there's no need to call
setPrevRev.
In preparation of running all tests inside container.
Currently, we run Jenkins in shared environment.
This is not good. Need manual Go runtime updates,
cannot run two different branches, port conflicts,
out of disk errors, etc.
Signed-off-by: Gyu-Ho Lee <gyuhox@gmail.com>
This commit adds an authentication mechanism to inter peer connection
(rafthttp). If the cert based peer auth is enabled and a new option
`--peer-cert-allowed-cn` is passed, an etcd process denies a peer
connection whose CN doesn't match.
With slow CPU, gRPC can lag behind with RPCs being sent before
calling 'Up', returning 'no address available' on the first try.
Signed-off-by: Gyu-Ho Lee <gyuhox@gmail.com>
Like the previous commit 10f783efdd12, this commit lets grpcproxy
forward an auth token supplied by its client in an explicit
manner. snapshot is a stream RPC so this process is required like
watch.
This commit lets grpcproxy handle authed watch. The main changes are:
1. forwrading a token of a new broadcast client
2. checking permission of a new client that participates to an
existing broadcast
Major updates to ugorji/go changed the signature of some
methods, resulting in the build failing for etcd/client
with default installation of the codec.
We regenerate the sources using codecgen with the new version
to reflect on the new changes.
Fixes#8573
Signed-off-by: Alexandre Beslic <abeslic@abronan.com>
Current `etcdctl role grant-permission` doesn't handle an empty key
("") correctly. Because the range permissions are treated as
BytesAffineInterval internally, just specifying the empty key as a
beginning of range introduces an invalid permission which doesn't work
and betray users' intuition. This commit fix the way of handling empty
key as a prefix or from key in permission granting.
Fix https://github.com/coreos/etcd/issues/8494
The logical input to Compare would be a LeaseID (type int64) but the
check panics if we give a LeaseID directly. Allow both so that we don't
unnecessarily annoy and confuse the programmer using the API in the most
logical way.
The golang/time package tracks monotonic time in each time.Time
returned by time.Now function for Go 1.9.
Use time.Time to measure whether a lease is expired and remove
previous pkg/monotime. Use zero time.Time to mean forever. No
expiration when expiry.IsZero() is true.
Fixes scripts and removes shellcheck warning suppressions.
* regexp warnings
* use ./*glob* so names don't become options
* use $(..) instead of legacy `..`
* read with -r to avoid mangling backslashes
* double quote to prevent globbing and word splitting
a. add comment of reopening file in cut function.
b. add const frameSizeBytes in decoder.
c. return directly if locked files empty in ReleaseLockTo function.
TestRecvMsgPreVote was intended to be introduced in
github.com/coreos/etcd/pull/6624 but was uncapitalized (search for
testRecvMsgPreVote instead) and then subsequently removed due to it
being unused.
Setting the ETCDCTL_API=3, then calling etcdctl was unwieldy and not
thread safe; all ctl v3 tests had to go through the ctlv3 wrapper and
could not easily mix with v2 commands.
If Close() is called before Cancel()'s cancel() completes, the
watch channel will be closed while the watch is still in the
synced list. If there's an event, etcd will try to write to a
closed channel. Instead, remove the watch from the bookkeeping
structures only after cancel completes, so Close() will always
call it.
Fixes#8443
Current error paths of TestV3WatchFromCurrentRevision don't clean the
used resources including goroutines. Because go's tests are executed
continuously in a single process, the leaked goroutines makes error
logs bloated like the below case:
https://jenkins-etcd-public.prod.coreos.systems/job/etcd-coverage/2143/
This commit lets the error paths clean the resources.
* flag: improve StringFlags by support set default value when init
when init flagSet, set default value should be moved to StringFlags init
func, which is more friendly
personal proposal
* flag: code improved for StringFlags
The stricter warnings on pkg/flags generates extra output that
break coverage tests. Unset the ETCDCTL_ARGS environment variable
so the warnings aren't printed.
grpc 1.3 uses MaxMsgSize() to limit received message size. However, grpc 1.4 introduces a 4mb default limit on send message size. In etcd, server shouldn't be limit size of message that it can be sent. Hence, set maximum size of send message using MaxSendMsgSize().
The penalty for TLS is non-trivial with race detection enabled.
Weakening the test certs from 4096-bit RSA to 2048-bit gives ~4x faster
runtimes for TestDoubleTLSClusterSizeOf3.
Update interacting_v3.md
Making it clear to the user that keys created via the v2 API are not readable by the etcdctl with the v3 API. A etcdctl v3 get of a v2 key key exits with 0 and no data, which is quite confusing, hopefully this just makes that it a bit clearer if the user upgraded etcd 3 in the past (and forget some of the 2.3 to 3.0 to 3.1 to 3.2 upgrade details) but never updated the API they used as v2 was the default and happen to trying to figure out wtf, this is a further reminder of that backward incompatibility.
Adding retry to acquire on failure causes Get to now retry until a
connection can be reestablished to the etcd server, causing the
timeout to trigger and fail the test.
Gets should retry on transient failure, but the txn inserts a write, skipping
the retry logic in the client. Instead, check the error if the txn should be
retried.
Fixes#8372
Causes etcdctl to hang with pending SIGQUIT signals according to
/proc/pid/status. The debugging wasn't very useful on travis
either; just totally remove it to get CI working again.
MacOS base64 uses -D and linux uses -d, while --decode
works on both platforms. And add missing server-ca-csr-wildcard.json.
Signed-off-by: Gyu-Ho Lee <gyuhox@gmail.com>
This pr changes UnsafeForEach to traverse on boltdb before on the buffer.
This ordering guarantees that UnsafeForEach traverses in the same order
before or after the commit of buffer.
Was defaulting to PeerTLSInfo for client connections to the etcd cluster.
Since proxy users may rely on this behavior, only use the client tls
info if given, and fall back to peer tls otherwise.
TestNodeWithSmallerTermCanCompleteElection tests the scenario where a
node that has been partitioned away (and fallen behind) rejoins the
cluster at about the same time the leader node gets partitioned away.
Previously the cluster would come to a standstill when run with PreVote
enabled.
When responding to Msg{Pre,}Vote messages we now include the term from
the message, not the local term. To see why consider the case where a
single node was previously partitioned away and it's local term is now
of date. If we include the local term (recall that for pre-votes we
don't update the local term), the (pre-)campaigning node on the other
end will proceed to ignore the message (it ignores all out of date
messages).
The term in the original message and current local term are the same in
the case of regular votes, but different for pre-votes.
NB: Had to change TestRecvMsgVote to include pb.Message.Term when
sending MsgVote messages. The new sanity checks on MsgVoteResp
(m.Term != 0) would panic with the old test as raft.Term would be equal
to 0 when responding with MsgVoteResp messages.
When digging into etcd/boltdb "storage space exceeded" issues, this metric may help answer questions about if/when compactions occured and how much data was freed.
Leak detector is catching goroutines trying to close files which appear
runtime related:
1 instances of:
syscall.Syscall(...)
/usr/local/golang/1.8.3/go/src/syscall/asm_linux_386.s:20 +0x5
syscall.Close(...)
/usr/local/golang/1.8.3/go/src/syscall/zsyscall_linux_386.go:296 +0x3d
os.(*file).close(...)
/usr/local/golang/1.8.3/go/src/os/file_unix.go:140 +0x62
It's unlikely a user goroutine will leak on file close; whitelist it.
Both grpc.Server.Stop and grpc.Server.GracefulStop close the listeners
first, to stop accepting the new connections. GracefulStop blocks until
all clients close their open transports(connections). Unary RPCs
only take a few seconds to finish. Stream RPCs, like watch, might never
close the connections from client side, thus making gRPC server wait
forever.
This patch still calls GracefulStop, but waits up to 10s before manually
closing the open transports.
Address https://github.com/coreos/etcd/issues/8224.
Signed-off-by: Gyu-Ho Lee <gyuhox@gmail.com>
Calling a WaitGroup.Done() in a defer will sometimes trigger the leak
detector since the WaitGroup.Wait() will unblock before the defer
block completes. If the leak detector runs before the Done() is
rescheduled, it will spuriously report the finishing Done() as a leak.
This happens enough in CI to be irritating; whitelist it and ignore.
From go-grpc v1.2.0, the number of max streams per client is set to 100
by default by the server side. This change makes it impossible
for third party proxies and custom clients to establish many streams.
Setting only latency options is a pain since every fault must
be disabled on the command line. Instead, by default start
as a standard bridge without any fault injection.
Since we implemented docs versioning, the default url is
https://cockroachlabs.com/docs/stable instead of
https://cockroachlabs.com/docs. We have a redirect in place
from /docs to /docs/stable, so existing links aren't broken,
but it's a better user experience to bypass the redirect.
The IP SAN check would always do a DNS SAN check if DNS is given
and the connection's IP is verified. Instead, don't check DNS
entries if there's a matching iP.
Fixes#8206
Use atomic functions to manipulate `rev` of `fakeRevGetter`
so that the tester goroutine can update the
value without race with the compactor's goroutine.
Current UserGet() and RoleGet() RPCs require admin permission. It
means that users cannot know which roles they belong to and what
permissions the roles have. This commit change the semantics and now
users can know their roles and permissions.
Default stm isolation level is serializable snapshot isolation, which
is different than snapshot isolation (SI)
Signed-off-by: Hui Kang <kangh@us.ibm.com>
Semaphore is failing with context exceeded errors and dial timeouts, only
returning an "Error: ..." from expect on etcdctl. So, only test for
"Error:" instead of grpc internal errors.
Instead of unconditionally randomizing, extend leases on promotion
if too many leases expire within the same time span. If the server
has few leases or spread out expires, there will be no extension.
Relying on mvcc to set the db size metric can cause it to
miss size changes when a txn commits after the last write
completes before a quiescent period. Instead, load the
db size on demand.
Fixes#8146
This commit add a new test case which ensures that non authorized RPCs
failed with ErrUserEmpty. The case can happen in a schedule like
below:
1. create a cluster
2. create clients
3. enable authentication of the cluster
4. the clients issue RPCs
Fix https://github.com/coreos/etcd/issues/7770
I see CI is failing to download release binaries
but exit code doesn't trigger CI job failure.
We need 'FAIL' string.
Signed-off-by: Gyu-Ho Lee <gyuhox@gmail.com>
Problem Observed
----------------
When there is no etcd process behind the proxy,
clients repeat resending lease grant requests without delay.
This behavior can cause abnormal resource consumption on CPU/RAM and
network.
Problem Detail
--------------
`LeaseGrant()` uses a bare protobuf client to forward requests.
However, it doesn't use `grpc.FailFast(false)`, which means the method returns
an `Unavailable` error immediately when no etcd process is available.
In clientv3, `Unavailable` errors are not considered the "Halt" error,
and library retries the request without delay.
Both clients and the proxy consume much CPU cycles to process retry requests.
Resolution
----------
Add `grpc.FailFast(false))` to `LeaseGrant()` of the `leaseProxy`.
This makes the proxy not to return immediately when no etcd process is
available. Clients will simply timeout requests instead.
This fixes failed RPC rate query, where we do not need
subtraction because we already query by the status code.
Also adds grpc_method to make it more specific. Most of the
time, the failure recovers within 10-second, which is our
Prometheus scrap interval, so 'rate' query might not cover
that time window, showing as 0s, but still shows up in the graph.
Signed-off-by: Gyu-Ho Lee <gyuhox@gmail.com>
The uncontended path for a mutex would fetch the minimum
revision key on the prefix after creating its entry in
the wait list. This fetch can be rolled into the txn for
creating the wait key, eliminating a round-trip for immediately
acquiring the lock.
The "too slow" comment is rather vague. If the server closes
the watch for being too slow (it doesn't seem to any more), the
watch client should gracefully resume instead of forcing the
user to handle it.
Also removed the 'opts' comment since it wasn't being maintained.
Save the snapshot index to the WAL before saving the snapshot to the
filesystem. This ensures that we'll only ever call wal.Open with a
snapshot that was previously saved to the WAL.
The old error was not clear about what URLs needed to be added, sometimes
truncating the list. To make it clearer, print out the missing entries
for --initial-cluster and print the full list of initial advertise peers.
Fixes#8079 and #7927
A read-only txn isn't serialized by raft, but it uses a fresh
read txn for every mvcc access prior to executing its request ops.
If a write txn modifies the keys matching the read txn's comparisons,
the read txn may return inconsistent results.
To fix, use the same read-only mvcc txn for the duration of the etcd
txn. Probably gets a modest txn speedup as well since there are
fewer read txn allocations.
This reverts commit 2bb33181b6. python-etcd
seems to depend on /v2/machines and the maintainer vanished. Plus, it is
prefixed with /v2/ so it probably can't be deprecated anyway.
Currently clients can revoke any lease without permission. This commit
lets etcdserver protect revoking with write permission.
This commit adds a mechanism for generating internal token. It is used
for indicating that LeaseRevoke was issued internally so it should be
able to delete any attached keys.
Current tests don't normally trigger the watch victim path because the
constants are too large; set the constants to small values and hammer
the store to cause watch delivery delays.
Was iterating over every file, reloading everything. Instead,
analyze the package directories. On my machine, the time for
vet checking goes from 34s to 3s. Scans more code too.
GetUser doesn't go through quorum, so issuing a user get to any member
of a cluster may fetch stale data from a slow member. Instead, use a
single member cluster for the test.
Fixes#7993
etcdctl was getting ctx errors from timing out trying to issue RPCs to
a TLS endpoint but without using TLS for transmission. Client should
immediately bail out with a time out error.
Dialing out without specifying TLS creds but giving https uses some
default behavior that depends on passing an endpoint with https to
Dial(), so it's not enough to completely rely on the balancer to supply
endpoints.
Fixes#8008
Also ctx-izes grpc.Dial
There's a workaround by running -run=Test but this periodically
comes up as an issue, so have `go test` only run Test* to stem
the complaints.
Fixes#8000
This commit adds a new option --txn-ops to `benchmark mvcc put`. A
number specified with this option will be used as a number of written
keys in a single transaction. It will be useful for checking the
effect of the batching.
The lock command is clumsy to use from the command line, needing mkfifo,
wait, etc. Instead, make like consul and support launching a command if
one is given.
boltdb on windows allocates a file with the full mmap size even if the
db is empty. Force the initial mmap size to 0 so there's no huge initial
db file on windows.
Fixes#7910
The go runtime won't always reinstall the default signal handler on the
SIGTERM path, so it's possible the signal won't terminate the process.
Instead, force SIG_DFL for the signal.
Persistent data should be configured in agent side.
There is no need to specify the data-dir in tester side.
Signed-off-by: Gyu-Ho Lee <gyuhox@gmail.com>
Computing the snapshot file path is error prone; snapshot recovery was
constructing file paths missing a path separator so the snapshot
would never be loaded. Instead, refactor the backend path handling
to use helper functions where possible.
Loading all keys at once would cause etcd to use twice as much
memory than it would need to serve the keys, causing RSS to spike on
boot. Instead, load the keys into the mvcc by chunk. Uses pipelining
for some concurrency.
Fixes#7822
election runner can deadlock in atomic release().
suppose election runner has two clients A and B.
if A is a leader and B is a follower, B obtains lock
for release() and waits for A to close(nextc) which signal
next round is ready. However, A can only close(nextc) if it
obtains lock for release(); hence deadlock.
this pr removes atomicity of validate() and release() in global.go
and gives the responsibility of locking to each runner.
FIXES#7891
In the case that follower recieves a snapshot from leader
and crashes before renaming xxx.snap.db to db but after
snapshot has persisted to .wal and .snap, restarting
follower results loading old db, new .wal, and new .snap.
This will causes a index mismatch between snap metadata index
and consistent index from db.
This pr forces an ordering where saving/renaming db must
happen after snapshot is persisted to wal and snap file.
this guarantees wal and snap files are newer than db.
on server restart, etcd server checks if snap index > db consistent index.
if yes, etcd server attempts to load xxx.snap.db where xxx=snap index
if there is any and panic other wise.
FIXES#7628
Trying to decouple the v2 client from SRV code. Can't move
into discovery/ since that creates a circular dependency. So,
give up and move all the SRV code into a new package.
This test verifies that adding a node does not cause the leader to step
down until at least one full ElectionTick cycle elapses.
Signed-off-by: Aaron Lehmann <aaron.lehmann@docker.com>
500ms keepalive delay on proxy side causes client to sometimes send
a second keepalive since it waits more than 500ms for the first response.
Fixes#7658
Connection pausing added another exit condition in the listener
path, causing the bridge to leak connections instead of closing
them when signalled to close. Also adds some additional Close
paranoia.
Fixes#7823
If the balancer update notification loop starts with a downed
connection and endpoints are switched while the old connection is up,
the balancer can potentially wait forever for an up connection without
refreshing the connections to reflect the current endpoints.
Instead, fetch upc/downc together, only caring about a single transition
either from down->up or up->down for each iteration
Simple way to reproduce failures: add time.Sleep(time.Second) to the
beginning of the update notification loop.
previously, apply() doesn't set consistIndex for EntryConfChange type.
this causes a misalignment between consistIndex and applied index
where EntryConfChange entry results setting applied index but not consistIndex.
suppose that addMember() is called and leader reflects that change.
1. applied index and consistIndex is now misaligned.
2. a new follower node joined.
3. leader sends the snapshot to follower
where the applied index is the snapshot metadata index.
4. follower node saves the snapshot and database(includes consistIndex) from leader.
5. restarting follower loads snapshot and database.
6. follower checks snapshot metadata index(same as applied index) and database consistIndex,
finds them don't match, and then panic.
FIXES#7834
If 'StartEtcd' returns before starting gRPC server
(e.g. mismatch snapshot, misconfiguration),
receiving from grpcServerC blocks forever. This patch
just closes the channel to not block on grpcServerC,
and proceeds to next stop operations in Close.
This was masking the issues in https://github.com/coreos/etcd/issues/7834
Signed-off-by: Gyu-Ho Lee <gyuhox@gmail.com>
This changes the baseConfig used when creating tls Configs to utilize
the GetCertificate and GetClientCertificate functions to always reload
the certificates from disk whenever they are needed.
Always reloading the certificates allows changing the certificates via
an external process without interrupting etcd.
Fixes#7576
Cherry-picked by Gyu-Ho Lee <gyuhox@gmail.com>
Original commit can be found at https://github.com/coreos/etcd/pull/7784
I found that enabling the CheckQuorum flag led to spurious leader
elections when new nodes joined. It looks like in the time between a new
node joining the cluster, and that node first communicating with the
leader, the quorum check could fail because the new node looks inactive.
To solve this, set the RecentActive flag when nodes are first added.
This gives a grace period for the node to communicate before it causes
the quorum check to fail.
Signed-off-by: Aaron Lehmann <aaron.lehmann@docker.com>
Watching from the leader's ModRevision could cause live-locking on
observe retry loops when the ModRevision is less than the compacted
revision. Instead, start watching the leader from at least the store
revision of the linearized read used to detect the current leader.
Fixes#7815
Printing the values in ctx.String() will data race if the value
is mutable and doesn't implement String(), which seems to be common.
Instead, just return a fixed string instead of computing it; v3client
watches don't need as much flexibility for creating separate strings,
so separate ctx strings probably aren't necessary at this point.
Fixes#7811
etcd tester runs etcd runner as a separate binary.
it signals sigstop to the runner when tester wants to stop stressing.
it signals sigcont to the runner when tester wants to start stressing.
when tester needs to clean up, it signals sigint to runner.
FIXES#7026
Problem is:
`Step1`: `etcdserver/raft.go`'s `Ready` process routine sends config-change entries via `r.applyc <- ap` (https://github.com/coreos/etcd/blob/master/etcdserver/raft.go#L193-L203)
`Step2`: `etcdserver/server.go`'s `*EtcdServer.run` routine receives this via `ap := <-s.r.apply()` (https://github.com/coreos/etcd/blob/master/etcdserver/server.go#L735-L738)
`StepA`: `Step1` proceeds without sync, right after sending `r.applyc <- ap`.
`StepB`: `Step2` proceeds without sync, right after `sched.Schedule(s.applyAll(&ep,&ap))`.
`StepC`: `etcdserver` tries to sync with `s.applyAll(&ep,&ap)` by calling `rh.waitForApply()`.
`rh.waitForApply()` waits for all pending jobs to finish in `pkg/schedule`
side. However, the order of `StepA`,`StepB`,`StepC` is not guaranteed. It
is possible that `StepC` happens first, and proceeds without waiting on
apply. And the restarting member comes back as a leader in single-node
cluster, when there is no synchronization between apply-layer and
config-change Raft entry apply. Confirmed with more debugging lines below,
only reproducible with slow CPU VM (~2 vCPU).
```
~:24.005397 I | etcdserver: starting server... [version: 3.2.0+git, cluster version: to_be_decided]
~:24.011136 I | etcdserver: [DEBUG] 29b2d24047a277df waitForApply before
~:24.011194 I | etcdserver: [DEBUG] 29b2d24047a277df starts wait for 0 pending jobs
~:24.011234 I | etcdserver: [DEBUG] 29b2d24047a277df finished wait for 0 pending jobs (current pending 0)
~:24.011268 I | etcdserver: [DEBUG] 29b2d24047a277df waitForApply after
~:24.011348 I | etcdserver: [DEBUG] [0] 29b2d24047a277df is scheduling conf change on 29b2d24047a277df
~:24.011396 I | etcdserver: [DEBUG] [1] 29b2d24047a277df is scheduling conf change on 5edf80e32a334cf0
~:24.011437 I | etcdserver: [DEBUG] [2] 29b2d24047a277df is scheduling conf change on e32e31e76c8d2678
~:24.011477 I | etcdserver: [DEBUG] 29b2d24047a277df scheduled conf change on 29b2d24047a277df
~:24.011509 I | etcdserver: [DEBUG] 29b2d24047a277df scheduled conf change on 5edf80e32a334cf0
~:24.011545 I | etcdserver: [DEBUG] 29b2d24047a277df scheduled conf change on e32e31e76c8d2678
~:24.012500 I | etcdserver: [DEBUG] 29b2d24047a277df applyConfChange on 29b2d24047a277df before
~:24.013014 I | etcdserver/membership: added member 29b2d24047a277df [unix://127.0.0.1:2100515039] to cluster 9250d4ae34216949
~:24.013066 I | etcdserver: [DEBUG] 29b2d24047a277df applyConfChange on 29b2d24047a277df after
~:24.013113 I | etcdserver: [DEBUG] 29b2d24047a277df applyConfChange on 29b2d24047a277df after trigger
~:24.013158 I | etcdserver: [DEBUG] 29b2d24047a277df applyConfChange on 5edf80e32a334cf0 before
~:24.013666 W | etcdserver: failed to send out heartbeat on time (exceeded the 10ms timeout for 11.964739ms)
~:24.013709 W | etcdserver: server is likely overloaded
~:24.013750 W | etcdserver: failed to send out heartbeat on time (exceeded the 10ms timeout for 12.057265ms)
~:24.013775 W | etcdserver: server is likely overloaded
~:24.013950 I | raft: 29b2d24047a277df is starting a new election at term 4
~:24.014012 I | raft: 29b2d24047a277df became candidate at term 5
~:24.014051 I | raft: 29b2d24047a277df received MsgVoteResp from 29b2d24047a277df at term 5
~:24.014107 I | raft: 29b2d24047a277df became leader at term 5
~:24.014146 I | raft: raft.node: 29b2d24047a277df elected leader 29b2d24047a277df at term 5
```
I am printing out the number of pending jobs before we call
`sched.WaitFinish(0)`, and there was no pending jobs, so it returned
immediately (before we schedule `applyAll`).
This is the root cause to:
- https://github.com/coreos/etcd/issues/7595
- https://github.com/coreos/etcd/issues/7739
- https://github.com/coreos/etcd/issues/7802
`sched.WaitFinish(0)` doesn't work when `len(f.pendings)==0` and
`f.finished==0`. Config-change is the first job to apply, so
`f.finished` is 0 in this case.
`f.finished` monotonically increases, so we need `WaitFinish(finished+1)`.
And `finished` must be the one before calling `Schedule`. This is safe
because `Schedule(applyAll)` is the only place adding jobs to `sched`.
Then scheduler waits on the single job of `applyAll`, by getting the
current number of finished jobs before sending `Schedule`.
Or just make it be blocked until `applyAll` routine triggers on the
config-change job. This patch just removes `waitForApply`, and
signal `raftDone` to wait until `applyAll` finishes applying entries.
Confirmed that it fixes the issue, as below:
```
~:43.198354 I | rafthttp: started streaming with peer 36cda5222aba364b (stream MsgApp v2 reader)
~:43.198740 I | etcdserver: [DEBUG] 3988bc20c2b2e40c waitForApply before
~:43.198836 I | etcdserver: [DEBUG] 3988bc20c2b2e40c starts wait for 0 pending jobs, 1 finished jobs
~:43.200696 I | integration: launched 3169361310155633349 ()
~:43.201784 I | etcdserver: [DEBUG] [0] 3988bc20c2b2e40c is scheduling conf change on 36cda5222aba364b
~:43.201884 I | etcdserver: [DEBUG] [1] 3988bc20c2b2e40c is scheduling conf change on 3988bc20c2b2e40c
~:43.201965 I | etcdserver: [DEBUG] [2] 3988bc20c2b2e40c is scheduling conf change on cf5d6cbc2a121727
~:43.202070 I | etcdserver: [DEBUG] 3988bc20c2b2e40c scheduled conf change on 36cda5222aba364b
~:43.202139 I | etcdserver: [DEBUG] 3988bc20c2b2e40c scheduled conf change on 3988bc20c2b2e40c
~:43.202204 I | etcdserver: [DEBUG] 3988bc20c2b2e40c scheduled conf change on cf5d6cbc2a121727
~:43.202444 I | etcdserver: [DEBUG] 3988bc20c2b2e40c applyConfChange on 36cda5222aba364b (request ID: 0) before
~:43.204486 I | etcdserver/membership: added member 36cda5222aba364b [unix://127.0.0.1:2100913646] to cluster 425d73f1b7b01674
~:43.204588 I | etcdserver: [DEBUG] 3988bc20c2b2e40c applyConfChange on 36cda5222aba364b (request ID: 0) after
~:43.204703 I | etcdserver: [DEBUG] 3988bc20c2b2e40c applyConfChange on 36cda5222aba364b (request ID: 0) after trigger
~:43.204791 I | etcdserver: [DEBUG] 3988bc20c2b2e40c applyConfChange on 3988bc20c2b2e40c (request ID: 0) before
~:43.205689 I | etcdserver/membership: added member 3988bc20c2b2e40c [unix://127.0.0.1:2101113646] to cluster 425d73f1b7b01674
~:43.205783 I | etcdserver: [DEBUG] 3988bc20c2b2e40c applyConfChange on 3988bc20c2b2e40c (request ID: 0) after
~:43.205929 I | etcdserver: [DEBUG] 3988bc20c2b2e40c applyConfChange on 3988bc20c2b2e40c (request ID: 0) after trigger
~:43.206056 I | etcdserver: [DEBUG] 3988bc20c2b2e40c applyConfChange on cf5d6cbc2a121727 (request ID: 0) before
~:43.207353 I | etcdserver/membership: added member cf5d6cbc2a121727 [unix://127.0.0.1:2100713646] to cluster 425d73f1b7b01674
~:43.207516 I | etcdserver: [DEBUG] 3988bc20c2b2e40c applyConfChange on cf5d6cbc2a121727 (request ID: 0) after
~:43.207619 I | etcdserver: [DEBUG] 3988bc20c2b2e40c applyConfChange on cf5d6cbc2a121727 (request ID: 0) after trigger
~:43.207710 I | etcdserver: [DEBUG] 3988bc20c2b2e40c finished scheduled conf change on 36cda5222aba364b
~:43.207781 I | etcdserver: [DEBUG] 3988bc20c2b2e40c finished scheduled conf change on 3988bc20c2b2e40c
~:43.207843 I | etcdserver: [DEBUG] 3988bc20c2b2e40c finished scheduled conf change on cf5d6cbc2a121727
~:43.207951 I | etcdserver: [DEBUG] 3988bc20c2b2e40c finished wait for 0 pending jobs (current pending 0, finished 1)
~:43.208029 I | rafthttp: started HTTP pipelining with peer cf5d6cbc2a121727
~:43.210339 I | rafthttp: peer 3988bc20c2b2e40c became active
~:43.210435 I | rafthttp: established a TCP streaming connection with peer 3988bc20c2b2e40c (stream MsgApp v2 reader)
~:43.210861 I | rafthttp: started streaming with peer 3988bc20c2b2e40c (writer)
~:43.211732 I | etcdserver: [DEBUG] 3988bc20c2b2e40c waitForApply after
```
Signed-off-by: Gyu-Ho Lee <gyuhox@gmail.com>
Turns out the optimization to ignore setting the init rev for
current revision watches breaks some ordering assumptions. Since
Watch only returns a channel once it gets a response, it should
bind the revision at the time of the first create response.
Was causing TestWatchReconnInit to fail.
Weaken TestV3ElectionObserve so it only checks that it observes a strictly
monotonically ascending leader transition sequence following the first
observed leader. First, the Observe will issue the leader channel before
getting a response for its first get; the election revision is only bound
after returning the channel. So, Observe can't be expected to always
return the leader at the time it was started. Second, Observe fetches
the current leader based on its create revision, but begins watching on its
ModRevision; this is important so that elections still work in case the
leader issues proclamations following a compaction that exceeds its
creation revision. So, Observe can't be expected to return the entire
proclamation sequence for a single leader.
Fixes#7749
o Set -e to abort script if a command fails.
o Allow custom docker 'TAG' from the environment.
o Move arch suffix to version to allow all images to
be put into a single repository.
o Enable cross builds. When doing cross builds where the
host and target architectures are different 'RUN mkdir'
will fail since the target container cannot be run on
the host. To work around this, create the directories
in build-docker, then use ADD in the Dockerfile.
o Add Dockerfile-release.arm64
Signed-off-by: Geoff Levand <geoff@infradead.org>
Uses GOARCH to build for a targeted arch.
Usage: GOARCH=... BINARYDIR=... BUILDDIR=... ./scripts/build-aci version
Signed-off-by: Geoff Levand <geoff@infradead.org>
Server Stop+Restart sometimes takes more than 500ms, so with a
one second window the lease client may not get a chance to issue
a keepalive and get a lease extension before the lease client
timer elapses. Instead, sleep for a shorter period of time (while
still guaranteeing a keepalive resend during quorum loss) and
skip the test if server restart takes longer than the lease TTL.
Fixes#7346
The current transport client TLS checking will pass an IP address into
VerifyHostnames if there is DNSNames SAN. However, the go runtime will
not resolve the DNS names to match the client IP. Intead, resolve the
names when checking.
raftNode was being initialized in start(), which was causing
hangs when trying to stop the etcd server since the stop channel
would not be initialized in time for the stop call. Instead,
setup non-configurable bits in a constructor.
Fixes#7668
Accumulation of old entries in the underlying array backing the
entries slice has been found to cause massive memory growth in
CockroachDB for workloads that do large (1MB) writes
(https://github.com/cockroachdb/cockroach/issues/14776)
This doesn't appear to have much consistent effect on the raft
benchmarks, although it's worth noting that they vary quite a bit
between runs so it's kind of tough to draw strong conclusions from them.
Let me know if there are any different benchmarks you'd like me to run!
Fixes#7746
benchmark old ns/op new ns/op delta
BenchmarkOneNode-8 3283 3125 -4.81%
benchmark old allocs new allocs delta
BenchmarkOneNode-8 6 6 +0.00%
benchmark old bytes new bytes delta
BenchmarkOneNode-8 796 727 -8.67%
benchmark old ns/op new ns/op delta
BenchmarkProposal3Nodes-8 4269 4337 +1.59%
benchmark old allocs new allocs delta
BenchmarkProposal3Nodes-8 15 13 -13.33%
benchmark old bytes new bytes delta
BenchmarkProposal3Nodes-8 5839 4544 -22.18%
Current benchmark picks destinations of RPCs in a random
manner. However, it will result divergent benchmarking result because
RPCs other than serializable range must be forwarded to a leader node
when a follower node receives it. This commit adds a new flag
--target-leader for avoid the problem. If the flag is passed,
benchmark always picks an endpoint of a leader node.
run goroutine was resetting a field for no reason and without holding a lock.
This patch cleans up the run goroutine management to make the start/stop path
less racey in general.
When apply-layer sees configuration change entry in
raft.Ready.CommittedEntries, the server should not proceed
until that entry is applied. Otherwise, follower's raft
layer advances, possibly election-timeouts, and becomes
the leader in single-node cluster, before add-node conf
change of other nodes is applied.
Fix https://github.com/coreos/etcd/issues/7595.
Signed-off-by: Gyu-Ho Lee <gyuhox@gmail.com>
Dual locking doesn't really give a convincing performance improvement and
the lock ordering makes it impossible to safely check if the TTL keeper
is enabled or not.
Fixes#7722
Pure Snapshot isolation would permit read conflicts. Change the name
from Snapshot to SerializableSnapshot to reflect that it will also
reject read conflicts.
There have been a few bug fixes in upstream.
Mainly for our grpc-go sub-dependencies
'idna' package introduces a new dependency 'golang.org/x/text'
Signed-off-by: Gyu-Ho Lee <gyuhox@gmail.com>
gRPC will replace empty strings with nil, but for the embedded case it's
possible for []byte{} to slip in and confuse the single key / >= key
watch logic.
The Get for the leader key will fetch based on the latest revision
instead of the deletion revision, missing leader updates between
the delete and the Get.
Although it's usually safe to skip these updates since they're
stale, it makes testing more difficult and in some cases the
full leader update history is desirable.
Addresses a case where two clients share the same lease. A client resigns but
disconnects / crashes and doesn't realize it. Another client reuses the
lease and gets leadership with a new key. The old client comes back and
tries to resign again, revoking the new leadership of the new client.
The full information about the leader's key is necessary to
safely use elections with transactions. Instead of returning
only the value on Leader(), return the entire GetResposne.
If gosimple or staticcheck had no output, it no other passes would be
applied because they were using `continue`. Similarly, the suppression
check never worked at all since it wasn't the result data into egrep.
Fixes#7685
Revoke expects the BatchTx lock to be held when holding the TxnDeleter
because it updates the lease bucket. The tests don't hold the lock so
it may race with the backend commit loop.
Fixes#7662
The coverage data is still useful even if some tests fail. Instead of
terminating the coverage pass on any test failure, collect and pass
the failed tests, generate the coverage report, then report the failed
packages and exit with an error.
etcd passes 'url.URL.Host' to 'SelfCert' which contains
client, peer port. 'net.ParseIP("127.0.0.1:2379")' returns
'nil', and the client on this self-cert will see errors
of '127.0.0.1 because it doesn't contain any IP SANs'
Signed-off-by: Gyu-Ho Lee <gyuhox@gmail.com>
This commit adds a new option --from-key to the command etcdctl role
grant-permission. If the option is passed, an open ended permission
will be granted to a role e.g. from start-key to any keys those are
larger than start-key.
Example:
$ ETCDCTL_API=3 bin/etcdctl --user root:p role grant r1 readwrite a b
$ ETCDCTL_API=3 bin/etcdctl --user root:p role grant --from-key r1 readwrite c
$ ETCDCTL_API=3 bin/etcdctl --user root:p role get r1
Role r1
KV Read:
[a, b) (prefix a)
[c, <open ended>
KV Write:
[a, b) (prefix a)
[c, <open ended>
Note that a closed parenthesis doesn't follow the above <open ended>
for indicating that the role has an open ended permission ("<open
ended>" is a valid range end).
Fixes https://github.com/coreos/etcd/issues/7468
Unconditionally opens a WithRequireLeader stream in the lease client. Any
keep alive channels opened using WithRequireLeader will be closed when
the leader is lost.
Fixes#7275
NewConfig() sets an initial cluster (potentially using a default name)
but we should clear it in the event another discovery option has been
specified.
PR #7517 attempted to address this however it only worked if the name
was left as "default".
(Completely) Fixes#7516
Perviously, we advance checkCompactionInterval more than we should.
The compaction might happen nondeterministically since there is no
synchronization before we call clock.Advance().
The number of rg.Wait() should be equal to the number of Advance() if
compactor routine and test routine run at the same pace. However, in our current
test, we call Advance() more than rg.Wait().
It works OK when the compactor routine runs "slower" than the test routine, which
is the common case. However, when the speed changes, the compactor routine might
block rg.Rev() since there is not enough calls of rg.Wait().
This commit forces the compactor and test routine to run at the same pace. And we supply
the exact number of Advance() and wg.Wait() that compactor needs.
Since the current revision is 0, it'll always be less than the compaction
revision. If the proxy sees a compaction, it would always reject the
current revision requests since it's less than the compaction revision.
Instead, check if the revision is historical before trying to reject on
compaction revision.
Fixes#7599
advance() should use rs.req.Entries[0].Data as the context instead of
req.Context for deletion. Since req.Context is never set, there won't be
any context being deleted from pendingReadIndex; results mem leak.
FIXES#7571
This commit change the type of cached permission information from the
home made thing to interval tree. It improves computational complexity
of permission checking from O(n) to O(lg n).
Fix https://github.com/coreos/etcd/issues/7526.
When resetting `bolt.Tx` in `defrag` and `batchTxBuffered.commit`
operation, we do not hold `readTx` lock, so the inflight range
requests can trigger panic in `mvcc.Range` paths. This fixes by
moving mutexes out and hold it while resetting the `readTx`.
Signed-off-by: Gyu-Ho Lee <gyuhox@gmail.com>
This commit resolves a TODO of auth store:
Current scheme of role deletion allows existing users to have the
deleted roles. Assume a case like below:
create a role r1
create a user u1 and grant r1 to u1
delete r1
After this sequence, u1 is still granted the role r1. So if admin
create a new role with the name r1, The new r1 is automatically
granted u1. In some cases, it would be confusing. So we need to
revoke the deleted role from all users.
For all intervals [x, y), Visit will visit intervals in ascending order
sorted by x. Also fixes a bug where Visit would not terminate the search
when requested by the visitor function.
When a grpc watch stream is torn down, it will join on its logical substream
goroutines by waiting for each to close a channel. This doesn't guarantee
the substream is fully exited, though, but only about to exit and can be
waiting to resume even after Watch.Close finishes. Instead, use a
waitgroup.Done at the very end of the substream defer.
Fixes#7573
It is almost same to Documentation/v2/authentication.md because a
major part of its user interface is shared with the v2 auth. The newly
added doc includes some refinements for the v3 auth.
Fix https://github.com/coreos/etcd/issues/7512.
If a server starts and aborts due to config error,
it is possible to get stuck in ReadyNotify waits.
This adds select case to get notified on stop channel.
Signed-off-by: Gyu-Ho Lee <gyuhox@gmail.com>
e2e tests use different invocations of etcdctl, so the endpoint used to get
the member list will not necessarily be the same to make the remove call.
Instead, select an endpoint that is not being remove, and connect with that.
Use the path/filepath package instead of the path package. The
path package assumes slash-separated paths, which doesn't work
on Windows. But path/filepath manipulates filename paths in a way
that's compatible across OSes.
Fix https://github.com/coreos/etcd/issues/7470.
This patch removes unnecessary term look-up in
'createMergedSnapshotMessage', which can trigger panic
if raft entry at etcdProgress.appliedi got compacted
by subsequent 'MsgSnap' messages--if a follower is
being (in this case, network latency spikes), it
could receive subsequent 'MsgSnap' requests from leader.
etcd server-side 'applyAll' routine and raft's Ready
processing routine becomes asynchronous after raft
entries are persisted. And given that raft Ready routine
takes less time to finish, it is possible that second
'MsgSnap' is being handled, while the slow 'applyAll'
is still processing the first(old) 'MsgSnap'. Then raft
Ready routine can compact the log entries at future
index to 'applyAll'. That is how 'createMergedSnapshotMessage'
tried to look up raft term with outdated etcdProgress.appliedi.
Signed-off-by: Gyu-Ho Lee <gyuhox@gmail.com>
In cases of multiple endpoints, it's possible member add would get a its
member list from a member that has not yet recognized the membership
update. Instead, confirm that the member list response is from the
member that acked the member add or from a member that has synced
with the cluster following the member add.
Fixes#7498
If substream is closing but outc is still open while reconnecting, then outc
would only be closed once the watch client would connect or once the watch
client is closed. This was leading to deadlocks in the proxy tests. Instead,
close immediately if the context is canceled.
Fixes#7503
AfterTest() has a delay that waits for runtime goroutines to exit;
CheckLeakedGoroutine does not. Since the test runner manages the
test cluster for examples, there is no delay between terminating
the cluster and checking for leaked goroutines. Instead, apply
Aftertest checking before running CheckLeakedGoroutine to let runtime
http goroutines finish.
If the context does not include auth information, get authinfo will
return a nil auth info and a nil error. This is then passed to
IsAdminPermitted, which would dereference the nil auth info.
ReadTxs are designed for read-only accesses to the backend using a
read-only boltDB transaction. Since BatchTx's are long-running
transactions, all writes to BatchTx will writeback to ReadTx, overlaying
the base read-only transaction.
This commit adds jwt token support in v3 auth API.
Remaining major ToDos:
- Currently token type isn't hidden from etcdserver. In the near
future the information should be completely invisible from
etcdserver package.
- Configurable expiration of token. Currently tokens can be valid
until keys are changed.
How to use:
1. generate keys for signing and verfying jwt tokens:
$ openssl genrsa -out app.rsa 1024
$ openssl rsa -in app.rsa -pubout > app.rsa.pub
2. add command line options to etcd like below:
--auth-token-type jwt \
--auth-jwt-pub-key app.rsa.pub --auth-jwt-priv-key app.rsa \
--auth-jwt-sign-method RS512
3. launch etcd cluster
Below is a performance comparison of serializable read w/ and w/o jwt
token. Every (3) etcd node is executed on a single machine. Signing
method is RS512 and key length is 1024 bit. As the results show, jwt
based token introduces a performance overhead but it would be
acceptable for a case that requires authentication.
w/o jwt token auth (no auth):
Summary:
Total: 1.6172 secs.
Slowest: 0.0125 secs.
Fastest: 0.0001 secs.
Average: 0.0002 secs.
Stddev: 0.0004 secs.
Requests/sec: 6183.5877
Response time histogram:
0.000 [1] |
0.001 [9982] |∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎
0.003 [1] |
0.004 [1] |
0.005 [0] |
0.006 [0] |
0.008 [6] |
0.009 [0] |
0.010 [1] |
0.011 [5] |
0.013 [3] |
Latency distribution:
10% in 0.0001 secs.
25% in 0.0001 secs.
50% in 0.0001 secs.
75% in 0.0001 secs.
90% in 0.0002 secs.
95% in 0.0002 secs.
99% in 0.0003 secs.
w/ jwt token auth:
Summary:
Total: 2.5364 secs.
Slowest: 0.0182 secs.
Fastest: 0.0002 secs.
Average: 0.0003 secs.
Stddev: 0.0005 secs.
Requests/sec: 3942.5185
Response time histogram:
0.000 [1] |
0.002 [9975] |∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎
0.004 [0] |
0.006 [1] |
0.007 [11] |
0.009 [2] |
0.011 [4] |
0.013 [5] |
0.015 [0] |
0.016 [0] |
0.018 [1] |
Latency distribution:
10% in 0.0002 secs.
25% in 0.0002 secs.
50% in 0.0002 secs.
75% in 0.0002 secs.
90% in 0.0003 secs.
95% in 0.0003 secs.
99% in 0.0004 secs.
Travis claimed errors of gosimple like below
(https://travis-ci.org/coreos/etcd/jobs/208098545):
gosimple checking failed:
contrib/raftexample/raftexample_test.go:78:6: should write erri := <-clus.errorC[i] instead of erri, _ := <-clus.errorC[i]
contrib/raftexample/raftexample_test.go:114:10: should write err := <-eC instead of err, _ := <-eC
This commit fixes the errors.
I want to create a more consistent naming system across the repos. Some
of our projects won't have libraries or tools (like Clair) but others
have integrated their software with Clair in various ways.
So, use a generic term: integrations.
TestNodeTick relies on a unreliable func `waitForSchedule` when running
with GOMAXPROCS > 1. This commit changes the test to make sure we stop
the node afte it drains the tick chan. The test should be reliable now.
The issue is caused by leader loss even after waitLeader() returns
which can happen if the test machine is flaky which triggers a leader loss
or the killed node is the leader since waitLeader() only scans followers in
TestRestartMember() and they can have the same older leader.
In those cases, clusterMustProgress() proceeds with no leader which triggers
the no leader error.
To get around that, use linearizable get in waitLeader() to ensure leader is up
and retries on kapi.create() in clusterMustProgress() to ensure it proceeds with
a leader.
FIX#7258
4a0f922 changed SelfCert to use a helper from pkg/fileutils which
introduced a transitive dependency on coreos/pkg/capnslog. This means
anyone who imports pkg/transport to use TLS with the clientv3 library
has the default stdlib logger hijacked by capnslog.
This PR reverts 4a0f922. There are no tests because 4a0f922 contained no
test and was not attached to a PR.
Fixes#7350
add and expose StopSignal to ExpectProcess allows user
to define what signal to send on ExpectProcess.close()
coverage testing code sets StopSignal to SIGTERM allowing
the test binary to shutdown gracefully so that it can generate
a coverage report.
Leadership timeout can sometimes take too long, such as in test cases.
However, it is possible to infer a leader is available based on RPCs
that must go through consensus. Therefore, have a way to update the
leadership status off the watch path.
No need of separate function to filter duplicates.
Just merge ranges in-place
```
go test -v -run=xxx -bench=BenchmarkMergeOld -benchmem
BenchmarkMergeOld-8 100000 13524 ns/op 1104 B/op 8 allocs/op
go test -v -run=xxx -bench=BenchmarkMergeNew -benchmem
BenchmarkMergeNew-8 100000 13432 ns/op 936 B/op 3 allocs/op
```
Not much performance boost, but less memory allocation
and simpler
Getting gosimple suggestion while running test script, so this PR is for fixing gosimple S1019 check.
raft/node_test.go:456:40: should use make([]raftpb.Entry, 1) instead (S1019)
raft/node_test.go:457:49: should use make([]raftpb.Entry, 1) instead (S1019)
raft/node_test.go:458:43: should use make([]raftpb.Message, 1) instead (S1019)
Refer https://github.com/dominikh/go-tools/blob/master/cmd/gosimple/README.md#checks for more information.
clientv3 integration test was using clientv3.NewKV, clientv3.NewWatcher, etc to create specific client.
replace those with direct client calls so that the direct calls can also test grpc proxy.
Because of my own silly mistake, current NewAuthStore() doesn't
initialize authStore in a correct manner. For example, after recovery
from snapshot, it cannot revive the flag of enabled/disabled. This
commit fixes the problem.
Fix https://github.com/coreos/etcd/issues/7165
This change is needed to handle process restarts with elections. When the
leader process is restarted, it should be able to hang on to the leadership
by using the existing lease.
Fixes#7166
It is not clear to users immediately what is the gRPC
gateway. Adding a more explaination to make it clear that
etcd3 supports HTTP API through the gateway.
Keep more wal entries in memory for fast follower recovery.
10,000 was a too small number that triggers quite a few snapshots.
ZK proves that 100,000 is a reasonable number for even old less prowerful
machines.
Eventually we should provide both count and max memory (for large entries).
When bootstrapping a cluster, the docker run command is mostly the same for all cluster member. This commits highlight the small variations between the commands to make them stand out.
2017-01-12 09:16:21 +01:00
1326 changed files with 173768 additions and 77426 deletions
A good bug report has some very specific qualities, so please read over our short document on
[reporting bugs][report_bugs] before you submit your bug report.
A good bug report has some very specific qualities, so please read over our short document on [reporting bugs][report_bugs] before submitting a bug report.
See [code changes](https://github.com/coreos/etcd/compare/v3.2.9...v3.2.10).
### Fixed
- Replace backend key-value database `boltdb/bolt` with [`coreos/bbolt`](https://github.com/coreos/bbolt/releases) to address [backend database size issue](https://github.com/coreos/etcd/issues/8009)
- Fix clientv3 balancer to handle [network partition](https://github.com/coreos/etcd/issues/8711)
- Upgrade [`google.golang.org/grpc`](https://github.com/grpc/grpc-go/releases) v1.2.1 to v1.7.3
- Upgrade [`github.com/grpc-ecosystem/grpc-gateway`](https://github.com/grpc-ecosystem/grpc-gateway/releases) v1.2 to v1.3
- Revert [discovery SRV auth `ServerName` with `*.{ROOT_DOMAIN}`](https://github.com/coreos/etcd/pull/8651) to support non-wildcard subject alternative names in the certs (see [issue #8445](https://github.com/coreos/etcd/issues/8445) for more contexts)
- For instance, `etcd --discovery-srv=etcd.local` will only authenticate peers/clients when the provided certs have root domain `etcd.local` (**not `*.etcd.local`**) as an entry in Subject Alternative Name (SAN) field
See [code changes](https://github.com/coreos/etcd/compare/v3.2.8...v3.2.9).
### Fixed(Security)
- Compile with [Go 1.8.4](https://groups.google.com/d/msg/golang-nuts/sHfMg4gZNps/a-HDgDDDAAAJ)
- Update `golang.org/x/crypto/bcrypt` (See [golang/crypto@6c586e1](https://github.com/golang/crypto/commit/6c586e17d90a7d08bbbc4069984180dce3b04117) for more)
- Fix discovery SRV bootstrapping to [authenticate `ServerName` with `*.{ROOT_DOMAIN}`](https://github.com/coreos/etcd/pull/8651), in order to support sub-domain wildcard matching (see [issue #8445](https://github.com/coreos/etcd/issues/8445) for more contexts)
- For instance, `etcd --discovery-srv=etcd.local` will only authenticate peers/clients when the provided certs have root domain `*.etcd.local` as an entry in Subject Alternative Name (SAN) field
- Automatic leadership transfer when leader steps down
- etcd flags
-`--strict-reconfig-check` flag is set by default
- Add `--log-output` flag
- Add `--metrics` flag
- v3 client
- Add SetEndpoints method; update endpoints at runtime
- Add Sync method; auto-update endpoints at runtime
- Add Lease TimeToLive API; fetch lease information
- replace Config.Logger field with global logger
- Get API responses are sorted in ascending order by default
- v3 etcdctl
- Add lease timetolive command
- Add `--print-value-only` flag to get command
- Add `--dest-prefix` flag to make-mirror command
- command get responses are sorted in ascending order by default
-`recipes` now conform to sessions defined in clientv3/concurrency
- ACI has symlinks to `/usr/local/bin/etcd*`
- experimental gRPC proxy feature
### Changed
- Deprecated following gRPC metrics in favor of [go-grpc-prometheus](https://github.com/grpc-ecosystem/go-grpc-prometheus):
-`etcd_grpc_requests_total`
-`etcd_grpc_requests_failed_total`
-`etcd_grpc_active_streams`
-`etcd_grpc_unary_requests_duration_seconds`
- etcd uses default route IP if advertise URL is not given
- Cluster rejects removing members if quorum will be lost
- SRV records (e.g., infra1.example.com) must match the discovery domain (i.e., example.com) if no custom certificate authority is given
- TLSConfig ServerName is ignored with user-provided certificates for backwards compatibility; to be deprecated in 3.2
- For example, `etcd --discovery-srv=example.com` will only authenticate peers/clients when the provided certs have root domain `example.com` as an entry in Subject Alternative Name (SAN) field
- Discovery now has upper limit for waiting on retries
- Warn on binding listeners through domain names; to be deprecated in 3.2
etcd is Apache 2.0 licensed and accepts contributions via GitHub pull requests. This document outlines some of the conventions on commit message formatting, contact points for developers and other resources to make getting your contribution into etcd easier.
etcd is Apache 2.0 licensed and accepts contributions via GitHub pull requests. This document outlines some of the conventions on commit message formatting, contact points for developers, and other resources to help get contributions into etcd.
before you submit your bug report. This document might contain links known
issues, another good reason to take a look there, before reporting your bug.
Reporting bugs is one of the best ways to contribute. However, a good bug report has some very specific qualities, so please read over our short document on [reporting bugs](https://github.com/coreos/etcd/blob/master/Documentation/reporting_bugs.md) before submitting a bug report. This document might contain links to known issues, another good reason to take a look there before reporting a bug.
## Contribution flow
This is a rough outline of what a contributor's workflow looks like:
- Create a topic branch from where you want to base your work. This is usually master.
- Create a topic branch from where to base the contribution. This is usually master.
- Make commits of logical units.
- Make sure your commit messages are in the proper format (see below).
- Push your changes to a topic branch in your fork of the repository.
- Make sure commit messages are in the proper format (see below).
- Push changes in a topic branch to a personal fork of the repository.
- Submit a pull request to coreos/etcd.
-Your PR must receive a LGTM from two maintainers found in the MAINTAINERS file.
-The PR must receive a LGTM from two maintainers found in the MAINTAINERS file.
Thanks for your contributions!
Thanks for contributing!
### Code style
@ -48,8 +44,7 @@ the body of the commit should describe the why.
```
scripts: add the test-cluster command
this uses tmux to setup a test cluster that you can easily kill and
start for debugging.
this uses tmux to setup a test cluster that can easily be killed and started for debugging.
Fixes #38
```
@ -64,7 +59,4 @@ The format can be described more formally as follows:
<footer>
```
The first line is the subject and should be no longer than 70 characters, the
second line is always blank, and other lines should be wrapped at 80 characters.
This allows the message to be easier to read on GitHub as well as in various
git tools.
The first line is the subject and should be no longer than 70 characters, the second line is always blank, and other lines should be wrapped at 80 characters. This allows the message to be easier to read on GitHub as well as in various git tools.
@ -39,7 +39,7 @@ The length of key name is always 64 bytes, which is a reasonable length of avera
## Data Size Threshold
- When etcd reaches data size threshold, it may trigger leader election easily and drop part of proposals.
-At most cases, etcd cluster should work smoothly if it doesn't hit the threshold. If it doesn't work well due to insufficient resources, you need to decrease its data size.
-For most cases, the etcd cluster should work smoothly if it doesn't hit the threshold. If it doesn't work well due to insufficient resources, decrease its data size.
| value bytes | key number limitation | suggested data size threshold(MB) | consumed RSS(MB) |
* Backwards-compatible bug fixes should target the master branch and subsequently be ported to stable branches.
* Once the master branch is ready for release, it will be tagged and become the new stable branch.
The etcd team has adopted a *rolling release model* and supports one stable version of etcd.
The etcd team has adopted a *rolling release model* and supports two stable versions of etcd.
### Master branch
@ -21,6 +21,6 @@ Before the release of the next stable version, feature PRs will be frozen. We wi
All branches with prefix `release-` are considered _stable_ branches.
After every minor release (http://semver.org/), we will have a new stable branch for that release. We will keep fixing the backwards-compatible bugs for the latest stable release, but not previous releases. The _patch_ release, incorporating any bug fixes, will be once every two weeks, given any patches.
After every minor release (http://semver.org/), we will have a new stable branch for that release. We will keep fixing the backwards-compatible bugs for the latest two stable releases. A _patch_ release to each supported release branch, incorporating any bug fixes, will be once every two weeks, given any patches.
| Lock | LockRequest | LockResponse | Lock acquires a distributed shared lock on a given named lock. On success, it will return a unique key that exists so long as the lock is held by the caller. This key can be used in conjunction with transactions to safely ensure updates to etcd only occur while holding lock ownership. The lock is held until Unlock is called on the key or the lease associate with the owner expires. |
| Unlock | UnlockRequest | UnlockResponse | Unlock takes a key returned by Lock and releases the hold on lock. The next Lock caller waiting for the lock will then be woken up and given ownership of the lock. |
| name | name is the identifier for the distributed shared lock to be acquired. | bytes |
| lease | lease is the ID of the lease that will be attached to ownership of the lock. If the lease expires or is revoked and currently holds the lock, the lock is automatically released. Calls to Lock with the same lease will be treated as a single acquistion; locking twice with the same lease is a no-op. | int64 |
| key | key is a key that will exist on etcd for the duration that the Lock caller owns the lock. Users should not modify this key or the lock may exhibit undefined behavior. | bytes |
| Campaign | CampaignRequest | CampaignResponse | Campaign waits to acquire leadership in an election, returning a LeaderKey representing the leadership if successful. The LeaderKey can then be used to issue new values on the election, transactionally guard API requests on leadership still being held, and resign from the election. |
| Proclaim | ProclaimRequest | ProclaimResponse | Proclaim updates the leader's posted value with a new value. |
| Leader | LeaderRequest | LeaderResponse | Leader returns the current election proclamation, if any. |
| Observe | LeaderRequest | LeaderResponse | Observe streams election proclamations in-order as made by the election's elected leaders. |
| Resign | ResignRequest | ResignResponse | Resign releases election leadership so other campaigners may acquire leadership on the election. |
| name | name is the election's identifier for the campaign. | bytes |
| lease | lease is the ID of the lease attached to leadership of the election. If the lease expires or is revoked before resigning leadership, then the leadership is transferred to the next campaigner, if any. | int64 |
| value | value is the initial proclaimed value set when the campaigner wins the election. | bytes |
| name | name is the election identifier that correponds to the leadership key. | bytes |
| key | key is an opaque key representing the ownership of the election. If the key is deleted, then leadership is lost. | bytes |
| rev | rev is the creation revision of the key. It can be used to test for ownership of an election during transactions by testing the key's creation revision matches rev. | int64 |
| lease | lease is the lease ID of the election leader. | int64 |
| type | type is the kind of event. If type is a PUT, it indicates new data has been stored to the key. If type is a DELETE, it indicates the key was deleted. | EventType |
| kv | kv holds the KeyValue for the event. A PUT event contains current kv pair. A PUT event with kv.Version=1 indicates the creation of a key. A DELETE/EXPIRE event contains the deleted key with its modification revision set to the revision of deletion. | KeyValue |
| prev_kv | prev_kv holds the key-value pair before the event happens. | KeyValue |
##### message `KeyValue` (mvcc/mvccpb/kv.proto)
| Field | Description | Type |
| ----- | ----------- | ---- |
| key | key is the key in bytes. An empty key is not allowed. | bytes |
| create_revision | create_revision is the revision of last creation on this key. | int64 |
| mod_revision | mod_revision is the revision of last modification on this key. | int64 |
| version | version is the version of the key. A deletion resets the version to zero and any modification of the key increases its version. | int64 |
| value | value is the value held by the key, in bytes. | bytes |
| lease | lease is the ID of the lease that attached to key. When the attached lease expires, the key will be deleted. If lease is 0, then no lease is attached to the key. | int64 |
@ -6,9 +6,11 @@ etcd v3 uses [gRPC][grpc] for its messaging protocol. The etcd project includes
## Using grpc-gateway
The gateway accepts a [JSON mapping][json-mapping] for etcd's [protocol buffer][api-ref] message definitions. Note that `key` and `value` fields are defined as byte arrays and therefore must be base64 encoded in JSON.
The gateway accepts a [JSON mapping][json-mapping] for etcd's [protocol buffer][api-ref] message definitions. Note that `key` and `value` fields are defined as byte arrays and therefore must be base64 encoded in JSON. The following examples use `curl`, but any HTTP/JSON client should work all the same.
Use `curl` to put and get a key:
### Put and get keys
Use the `/v3beta/kv/range` and `/v3beta/kv/put` services to read and write keys:
| Range | RangeRequest | RangeResponse | Range gets the keys in the range from the key-value store. |
@ -60,6 +58,7 @@ for grpc-gateway
| LeaseRevoke | LeaseRevokeRequest | LeaseRevokeResponse | LeaseRevoke revokes a lease. All keys attached to the lease will expire and be deleted. |
| LeaseKeepAlive | LeaseKeepAliveRequest | LeaseKeepAliveResponse | LeaseKeepAlive keeps the lease alive by streaming keep alive requests from the client to the server and streaming keep alive responses from the server to the client. |
| Status | StatusRequest | StatusResponse | Status gets the status of the member. |
| Defragment | DefragmentRequest | DefragmentResponse | Defragment defragments a member's backend database to recover storage space. |
| Hash | HashRequest | HashResponse | Hash returns the hash of the local KV state for consistency checking purpose. This is designed for testing; do not use this in production when there are ongoing transactions. |
| Hash | HashRequest | HashResponse | Hash computes the hash of the KV's backend. This is designed for testing; do not use this in production when there are ongoing transactions. |
| HashKV | HashKVRequest | HashKVResponse | HashKV computes the hash of all MVCC keys up to a given revision. |
| Snapshot | SnapshotRequest | SnapshotResponse | Snapshot sends a snapshot of the entire backend from a member over a stream to a client. |
| MoveLeader | MoveLeaderRequest | MoveLeaderResponse | MoveLeader requests current leader node to transfer its leadership to transferee. |
default, used to query if any alarm is active space quota is exhausted
| Field | Description | Type |
| ----- | ----------- | ---- |
| action | action is the kind of alarm request to issue. The action may GET alarm statuses, ACTIVATE an alarm, or DEACTIVATE a raised alarm. | AlarmAction |
@ -405,6 +404,8 @@ CompactionRequest compacts the key-value store up to a given revision. All super
| create_revision | create_revision is the creation revision of the given key | int64 |
| mod_revision | mod_revision is the last modified revision of the given key. | int64 |
| value | value is the value of the given key, in bytes. | bytes |
| lease | lease is the lease id of the given key. | int64 |
| range_end | range_end compares the given target to all keys in the range [key, range_end). See RangeRequest for more details on key ranges. | bytes |
@ -427,8 +428,8 @@ Empty field.
| Field | Description | Type |
| ----- | ----------- | ---- |
| key | key is the first key to delete in the range. | bytes |
| range_end | range_end is the key following the last key to delete for the range [key, range_end). If range_end is not given, the range is defined to contain only the key argument. If range_end is one bit larger than the given key, then the range is all the all keys with the prefix (the given key). If range_end is '\0', the range is all keys greater than or equal to the key argument. | bytes |
| prev_kv | If prev_kv is set, etcd gets the previous key-value pairs before deleting it. The previous key-value pairs will be returned in the delte response. | bool |
| range_end | range_end is the key following the last key to delete for the range [key, range_end). If range_end is not given, the range is defined to contain only the key argument. If range_end is one bit larger than the given key, then the range is all the keys with the prefix (the given key). If range_end is '\0', the range is all keys greater than or equal to the key argument. | bytes |
| prev_kv | If prev_kv is set, etcd gets the previous key-value pairs before deleting it. The previous key-value pairs will be returned in the delete response. | bool |
| value | value is the value, in bytes, to associate with the key in the key-value store. | bytes |
| lease | lease is the lease ID to associate with the key in the key-value store. A lease value of 0 indicates no lease. | int64 |
| prev_kv | If prev_kv is set, etcd gets the previous key-value pair before changing it. The previous key-value pair will be returned in the put response. | bool |
| ignore_value | If ignore_value is set, etcd updates the key using its current value. Returns an error if the key does not exist. | bool |
| ignore_lease | If ignore_lease is set, etcd updates the key using its current lease. Returns an error if the key does not exist. | bool |
@ -632,9 +695,9 @@ Empty field.
| Field | Description | Type |
| ----- | ----------- | ---- |
| key | default, no sorting lowest target value first highest target value first key is the first key for the range. If range_end is not given, the request only looks up key. | bytes |
| range_end | range_end is the upper bound on the requested range [key, range_end). If range_end is '\0', the range is all keys >= key. If the range_end is one bit larger than the given key, then the range requests get the all keys with the prefix (the given key). If both key and range_end are '\0', then range requests returns all keys. | bytes |
| limit | limit is a limit on the number of keys returned for the request. | int64 |
| key | key is the first key for the range. If range_end is not given, the request only looks up key. | bytes |
| range_end | range_end is the upper bound on the requested range [key, range_end). If range_end is '\0', the range is all keys >= key. If range_end is key plus one (e.g., "aa"+1 == "ab", "a\xff"+1 == "b"), then the range request gets all keys prefixed with key. If both key and range_end are '\0', then the range request returns all keys. | bytes |
| limit | limit is a limit on the number of keys returned for the request. When limit is set to 0, it is treated as no limit. | int64 |
| revision | revision is the point-in-time of the key-value store to use for the range. If revision is less or equal to zero, the range is over the newest key-value store. If the revision has been compacted, ErrCompacted is returned as a response. | int64 |
| sort_order | sort_order is the order for returned sorted results. | SortOrder |
| sort_target | sort_target is the key-value field to use for sorting. | SortTarget |
@ -667,6 +730,7 @@ Empty field.
| request_range | | RangeRequest |
| request_put | | PutRequest |
| request_delete_range | | DeleteRangeRequest |
| request_txn | | TxnRequest |
@ -689,6 +753,7 @@ Empty field.
| response_range | | RangeResponse |
| response_put | | PutResponse |
| response_delete_range | | DeleteRangeResponse |
| response_txn | | TxnResponse |
@ -765,7 +830,7 @@ From google paxosdb paper: Our implementation hinges around a powerful primitive
| range_end | range_end is the end of the range [key, range_end) to watch. If range_end is not given, only the key argument is watched. If range_end is equal to '\0', all keys greater than or equal to the key argument are watched. If the range_end is one bit larger than the given key, then all keys with the prefix (the given key) will be watched. | bytes |
| start_revision | start_revision is an optional revision to watch from (inclusive). No start_revision is "now". | int64 |
| progress_notify | progress_notify is set so that the etcd server will periodically send a WatchResponse with no events to the new watcher if there are no recent events. It is useful when clients wish to recover a disconnected watcher starting from a recent known revision. The etcd server may decide how often it will send notifications based on current load. | bool |
| filters | filter out put event. filter out delete event. filters filter the events at server side before it sends back to the watcher. | (slice of) FilterType |
| filters | filters filter the events at server side before it sends back to the watcher. | (slice of) FilterType |
| prev_kv | If prev_kv is set, created watcher gets the previous KV before the event happens. If the previous KV is already compacted, nothing will be returned. | bool |
@ -789,6 +854,7 @@ From google paxosdb paper: Our implementation hinges around a powerful primitive
| created | created is set to true if the response is for a create watch request. The client should record the watch_id and expect to receive events for the created watcher from the same stream. All events sent to the created watcher will attach with the same watch_id. | bool |
| canceled | canceled is set to true if the response is for a cancel watch request. No further events will be sent to the canceled watcher. | bool |
| compact_revision | compact_revision is set to the minimum index if a watcher tries to watch at a compacted index. This happens when creating a watcher at a compacted revision or the watcher cannot catch up with the progress of the key-value store. The client should treat the watcher as canceled and should not try to create any watcher with the same start_revision again. | int64 |
| cancel_reason | cancel_reason indicates the reason for canceling the watcher. | string |
"summary":"Campaign waits to acquire leadership in an election, returning a LeaderKey\nrepresenting the leadership if successful. The LeaderKey can then be used\nto issue new values on the election, transactionally guard API requests on\nleadership still being held, and resign from the election.",
"summary":"Resign releases election leadership so other campaigners may acquire\nleadership on the election.",
"operationId":"Resign",
"responses":{
"200":{
"description":"",
"schema":{
"$ref":"#/definitions/v3electionpbResignResponse"
}
}
},
"parameters":[
{
"name":"body",
"in":"body",
"required":true,
"schema":{
"$ref":"#/definitions/v3electionpbResignRequest"
}
}
],
"tags":[
"Election"
]
}
}
},
"definitions":{
"etcdserverpbResponseHeader":{
"type":"object",
"properties":{
"cluster_id":{
"type":"string",
"format":"uint64",
"description":"cluster_id is the ID of the cluster which sent the response."
},
"member_id":{
"type":"string",
"format":"uint64",
"description":"member_id is the ID of the member which sent the response."
},
"revision":{
"type":"string",
"format":"int64",
"description":"revision is the key-value store revision when the request was applied."
},
"raft_term":{
"type":"string",
"format":"uint64",
"description":"raft_term is the raft term when the request was applied."
}
}
},
"mvccpbKeyValue":{
"type":"object",
"properties":{
"key":{
"type":"string",
"format":"byte",
"description":"key is the key in bytes. An empty key is not allowed."
},
"create_revision":{
"type":"string",
"format":"int64",
"description":"create_revision is the revision of last creation on this key."
},
"mod_revision":{
"type":"string",
"format":"int64",
"description":"mod_revision is the revision of last modification on this key."
},
"version":{
"type":"string",
"format":"int64",
"description":"version is the version of the key. A deletion resets\nthe version to zero and any modification of the key\nincreases its version."
},
"value":{
"type":"string",
"format":"byte",
"description":"value is the value held by the key, in bytes."
},
"lease":{
"type":"string",
"format":"int64",
"description":"lease is the ID of the lease that attached to key.\nWhen the attached lease expires, the key will be deleted.\nIf lease is 0, then no lease is attached to the key."
}
}
},
"v3electionpbCampaignRequest":{
"type":"object",
"properties":{
"name":{
"type":"string",
"format":"byte",
"description":"name is the election's identifier for the campaign."
},
"lease":{
"type":"string",
"format":"int64",
"description":"lease is the ID of the lease attached to leadership of the election. If the\nlease expires or is revoked before resigning leadership, then the\nleadership is transferred to the next campaigner, if any."
},
"value":{
"type":"string",
"format":"byte",
"description":"value is the initial proclaimed value set when the campaigner wins the\nelection."
}
}
},
"v3electionpbCampaignResponse":{
"type":"object",
"properties":{
"header":{
"$ref":"#/definitions/etcdserverpbResponseHeader"
},
"leader":{
"$ref":"#/definitions/v3electionpbLeaderKey",
"description":"leader describes the resources used for holding leadereship of the election."
}
}
},
"v3electionpbLeaderKey":{
"type":"object",
"properties":{
"name":{
"type":"string",
"format":"byte",
"description":"name is the election identifier that correponds to the leadership key."
},
"key":{
"type":"string",
"format":"byte",
"description":"key is an opaque key representing the ownership of the election. If the key\nis deleted, then leadership is lost."
},
"rev":{
"type":"string",
"format":"int64",
"description":"rev is the creation revision of the key. It can be used to test for ownership\nof an election during transactions by testing the key's creation revision\nmatches rev."
},
"lease":{
"type":"string",
"format":"int64",
"description":"lease is the lease ID of the election leader."
}
}
},
"v3electionpbLeaderRequest":{
"type":"object",
"properties":{
"name":{
"type":"string",
"format":"byte",
"description":"name is the election identifier for the leadership information."
}
}
},
"v3electionpbLeaderResponse":{
"type":"object",
"properties":{
"header":{
"$ref":"#/definitions/etcdserverpbResponseHeader"
},
"kv":{
"$ref":"#/definitions/mvccpbKeyValue",
"description":"kv is the key-value pair representing the latest leader update."
}
}
},
"v3electionpbProclaimRequest":{
"type":"object",
"properties":{
"leader":{
"$ref":"#/definitions/v3electionpbLeaderKey",
"description":"leader is the leadership hold on the election."
},
"value":{
"type":"string",
"format":"byte",
"description":"value is an update meant to overwrite the leader's current value."
}
}
},
"v3electionpbProclaimResponse":{
"type":"object",
"properties":{
"header":{
"$ref":"#/definitions/etcdserverpbResponseHeader"
}
}
},
"v3electionpbResignRequest":{
"type":"object",
"properties":{
"leader":{
"$ref":"#/definitions/v3electionpbLeaderKey",
"description":"leader is the leadership to relinquish by resignation."
"summary":"Lock acquires a distributed shared lock on a given named lock.\nOn success, it will return a unique key that exists so long as the\nlock is held by the caller. This key can be used in conjunction with\ntransactions to safely ensure updates to etcd only occur while holding\nlock ownership. The lock is held until Unlock is called on the key or the\nlease associate with the owner expires.",
"operationId":"Lock",
"responses":{
"200":{
"description":"",
"schema":{
"$ref":"#/definitions/v3lockpbLockResponse"
}
}
},
"parameters":[
{
"name":"body",
"in":"body",
"required":true,
"schema":{
"$ref":"#/definitions/v3lockpbLockRequest"
}
}
],
"tags":[
"Lock"
]
}
},
"/v3beta/lock/unlock":{
"post":{
"summary":"Unlock takes a key returned by Lock and releases the hold on lock. The\nnext Lock caller waiting for the lock will then be woken up and given\nownership of the lock.",
"operationId":"Unlock",
"responses":{
"200":{
"description":"",
"schema":{
"$ref":"#/definitions/v3lockpbUnlockResponse"
}
}
},
"parameters":[
{
"name":"body",
"in":"body",
"required":true,
"schema":{
"$ref":"#/definitions/v3lockpbUnlockRequest"
}
}
],
"tags":[
"Lock"
]
}
}
},
"definitions":{
"etcdserverpbResponseHeader":{
"type":"object",
"properties":{
"cluster_id":{
"type":"string",
"format":"uint64",
"description":"cluster_id is the ID of the cluster which sent the response."
},
"member_id":{
"type":"string",
"format":"uint64",
"description":"member_id is the ID of the member which sent the response."
},
"revision":{
"type":"string",
"format":"int64",
"description":"revision is the key-value store revision when the request was applied."
},
"raft_term":{
"type":"string",
"format":"uint64",
"description":"raft_term is the raft term when the request was applied."
}
}
},
"v3lockpbLockRequest":{
"type":"object",
"properties":{
"name":{
"type":"string",
"format":"byte",
"description":"name is the identifier for the distributed shared lock to be acquired."
},
"lease":{
"type":"string",
"format":"int64",
"description":"lease is the ID of the lease that will be attached to ownership of the\nlock. If the lease expires or is revoked and currently holds the lock,\nthe lock is automatically released. Calls to Lock with the same lease will\nbe treated as a single acquistion; locking twice with the same lease is a\nno-op."
}
}
},
"v3lockpbLockResponse":{
"type":"object",
"properties":{
"header":{
"$ref":"#/definitions/etcdserverpbResponseHeader"
},
"key":{
"type":"string",
"format":"byte",
"description":"key is a key that will exist on etcd for the duration that the Lock caller\nowns the lock. Users should not modify this key or the lock may exhibit\nundefined behavior."
}
}
},
"v3lockpbUnlockRequest":{
"type":"object",
"properties":{
"key":{
"type":"string",
"format":"byte",
"description":"key is the lock ownership key granted by Lock."
@ -4,8 +4,4 @@ For the most part, the etcd project is stable, but we are still moving fast! We
## The current experimental API/features are:
- [gateway][gateway]: beta, to be stable in 3.2 release
- [gRPC proxy][grpc-proxy]: alpha, to be stable in 3.2 release
[gateway]: ../op-guide/gateway.md
[grpc-proxy]: ../op-guide/grpc_proxy.md
- [KV ordering](https://godoc.org/github.com/coreos/etcd/clientv3/ordering) wrapper. When an etcd client switches endpoints, responses to serializable reads may go backward in time if the new endpoint is lagging behind the rest of the cluster. The ordering wrapper caches the current cluster revision from response headers. If a response revision is less than the cached revision, the client selects another endpoint and reissues the read. Enable in grpcproxy with `--experimental-serializable-ordering`.
Users mostly interact with etcd by putting or getting the value of a key. This section describes how to do that by using etcdctl, a command line tool for interacting with etcd server. The concepts described here should apply to the gRPC APIs or client library APIs.
By default, etcdctl talks to the etcd server with the v2 API for backward compatibility. For etcdctl to speak to etcd using the v3 API, the API version must be set to version 3 via the `ETCDCTL_API` environment variable.
By default, etcdctl talks to the etcd server with the v2 API for backward compatibility. For etcdctl to speak to etcd using the v3 API, the API version must be set to version 3 via the `ETCDCTL_API` environment variable. However note that any key that was created using the v2 API will not be able to be queried via the v3 API. A v3 API ```etcdctl get``` of a v2 key will exit with 0 and no key data, this is the expected behaviour.
```bash
export ETCDCTL_API=3
@ -10,7 +11,7 @@ export ETCDCTL_API=3
## Find versions
etcdctl version and Server API version can be useful in finding the appropriate commands to be used for performing various opertions on etcd.
etcdctl version and Server API version can be useful in finding the appropriate commands to be used for performing various operations on etcd.
Here is the command to find the versions:
@ -215,7 +216,7 @@ $ etcdctl del foo foo9
Here is the command to delete key `zoo` with the deleted key value pair returned:
```bash
$ etcdctl del --prev-kv zoo
$ etcdctl del --prev-kv zoo
1 # one key is deleted
zoo # deleted key
val # the value of the deleted key
@ -224,7 +225,7 @@ val # the value of the deleted key
Here is the command to delete keys having prefix as `zoo`:
```bash
$ etcdctl del --prefix zoo
$ etcdctl del --prefix zoo
2 # two keys are deleted
```
@ -290,7 +291,7 @@ barz1
Here is the command to watch on multiple keys `foo` and `zoo`:
```bash
$ etcdctl watch -i
$ etcdctl watch -i
$ watch foo
$ watch zoo
# in another terminal: etcdctl put foo bar
@ -430,9 +431,9 @@ Here is the command to keep the same lease alive:
etcd is designed to handle small key value pairs typical for metadata. Larger requests will work, but may increase the latency of other requests. For the time being, etcd guarantees to support RPC requests with up to 1MB of data. In the future, the size limit may be loosened or made it configurable.
etcd is designed to handle small key value pairs typical for metadata. Larger requests will work, but may increase the latency of other requests. For the time being, etcd guarantees to support RPC requests with up to 1MB of data. In the future, the size limit may be loosened or made configurable.
For testing and development deployments, the quickest and easiest way is to set up a local cluster. For a production deployment, refer to the [clustering][clustering] section.
For testing and development deployments, the quickest and easiest way is to configure a local cluster. For a production deployment, refer to the [clustering][clustering] section.
## Local standalone cluster
Deploying an etcd cluster as a standalone cluster is straightforward. Start it with just one command:
### Starting a cluster
Run the following to deploy an etcd cluster as a standalone cluster:
```
$ ./etcd
...
```
The started etcd member listens on `localhost:2379`for client requests.
If the `etcd` binary is not present in the current working directory, it might be located either at `$GOPATH/bin/etcd` or at `/usr/local/bin/etcd`. Run the command appropriately.
To interact with the started cluster by using etcdctl:
The running etcd member listens on `localhost:2379` for client requests.
```
# use API version 3
$ export ETCDCTL_API=3
### Interacting with the cluster
$ ./etcdctl put foo bar
OK
Use `etcdctl` to interact with the running cluster:
$ ./etcdctl get foo
bar
```
1. Configure the environment to have `ETCDCTL_API=3` so `etcdctl` uses the etcd API version 3 instead of defaulting to version 2.
```
# use API version 3
$ export ETCDCTL_API=3
```
2. Store an example key-value pair in the cluster:
```
$ ./etcdctl put foo bar
OK
```
If OK is printed, storing key-value pair is successful.
3. Retrieve the value of `foo`:
```
$ ./etcdctl get foo
bar
```
If `bar` is returned, interaction with the etcd cluster is working as expected.
## Local multi-member cluster
A `Procfile` at the base of this git repo is provided to easily set up a local multi-member cluster. To start a multi-member cluster go to the root of an etcd source tree and run:
### Starting a cluster
```
# install goreman program to control Profile-based applications.
$ go get github.com/mattn/goreman
$ goreman -f Procfile start
...
```
A `Procfile` at the base of the etcd git repository is provided to easily configure a local multi-member cluster. To start a multi-member cluster, navigate to the root of the etcd source tree and perform the following:
The started members listen on `localhost:2379`, `localhost:22379`, and `localhost:32379` for client requests respectively.
1. Install `goreman` to control Procfile-based applications:
To interact with the started cluster by using etcdctl:
```
$ go get github.com/mattn/goreman
```
```
# use API version 3
$ export ETCDCTL_API=3
2. Start a cluster with `goreman` using etcd's stock Procfile:
$ etcdctl --write-out=table --endpoints=localhost:2379 member list
Restarting the member re-establish the connection. `etcdctl` will now be able to retrieve the key successfully. To learn more about interacting with etcd, read [interacting with etcd section][interacting].
The guide talks about how to release a new version of etcd.
The procedure includes some manual steps for sanity checking but it can probably be further scripted. Please keep this document up-to-date if making changes to the release process.
The procedure includes some manual steps for sanity checking, but it can probably be further scripted. Please keep this document up-to-date if making changes to the release process.
## Prepare release
@ -25,8 +25,10 @@ All releases version numbers follow the format of [semantic versioning 2.0.0](ht
### Patch version release
-Discuss about commits that are backported to the patch release. The commits should not include merge commits.
-Cherry-pick these commits starting from the oldest one into stable branch.
-To request a backport, devlopers submit cherrypick PRs targeting the release branch. The commits should not include merge commits. The commits should be restricted to bug fixes and security patches.
-The cherrypick PRs should target the appropriate release branch (`base:release-<major>-<minor>`). `hack/patch/cherrypick.sh` may be used to automatically generate cherrypick PRs.
- The release patch manager reviews the cherrypick PRs. Please discuss carefully what is backported to the patch release. Each patch release should be strictly better than it's predecessor.
- The release patch manager will cherry-pick these commits starting from the oldest one into stable branch.
## Write release note
@ -53,12 +55,12 @@ All releases version numbers follow the format of [semantic versioning 2.0.0](ht
It generates all release binaries and images under directory ./release.
## Sign binaries and images
## Sign binaries, images, and source code
etcd project key must be used to sign the generated binaries and images.`$SUBKEYID` is the key ID of etcd project Yubikey. Connect the key and run `gpg2 --card-status` to get the ID.
@ -66,8 +68,17 @@ The following commands are used for public release sign:
```
cd release
for i in etcd-*{.zip,.tar.gz}; do gpg2 --default-key $SUBKEYID --armor --output ${i}.asc --detach-sign ${i}; done
for i in etcd-*{.zip,.tar.gz}; do gpg2 --verify ${i}.asc ${i}; done
for i in etcd-*{.zip,.tar.gz,.aci}; do gpg2 --default-key $SUBKEYID --armor --output ${i}.asc --detach-sign ${i}; done
for i in etcd-*{.zip,.tar.gz,.aci}; do gpg2 --verify ${i}.asc ${i}; done
- Add `latest` tag to the new image on [gcr.io](https://console.cloud.google.com/gcr/images/etcd-development/GLOBAL/etcd?project=etcd-development&authuser=1) if this is a stable release.
The etcd performance benchmarks run etcd on 8 vCPU, 16GB RAM, 50GB SSD GCE instances, but any relatively modern machine with low latency storage and a few gigabytes of memory should suffice for most use cases. Applications with large v2 data stores will require more memory than a large v3 data store since data is kept in anonymous memory instead of memory mapped from a file. than For running etcd on a cloud provider, we suggest at least a medium instance on AWS or a standard-1 instance on GCE.
The etcd performance benchmarks run etcd on 8 vCPU, 16GB RAM, 50GB SSD GCE instances, but any relatively modern machine with low latency storage and a few gigabytes of memory should suffice for most use cases. Applications with large v2 data stores will require more memory than a large v3 data store since data is kept in anonymous memory instead of memory mapped from a file. For running etcd on a cloud provider, see the [Example hardware configuration][example-hardware-configurations] documentation.
## Download the pre-built binary
@ -10,16 +10,7 @@ The easiest way to get etcd is to use one of the pre-built release binaries whic
## Build the latest version
For those wanting to try the very latest version, build etcd from the `master` branch. [Go](https://golang.org/) version 1.7+ is required to build the latest version of etcd. To ensure etcd is built against well-tested libraries, etcd vendors its dependencies for official release binaries. However, etcd's vendoring is also optional to avoid potential import conflicts when embedding the etcd server or using the etcd client.
First, confirm go 1.7+ is installed:
```sh
# go is required
$ go version
go version go1.7.3 darwin/amd64
```
For those wanting to try the very latest version, build etcd from the `master` branch. [Go](https://golang.org/) version 1.9+ is required to build the latest version of etcd. To ensure etcd is built against well-tested libraries, etcd vendors its dependencies for official release binaries. However, etcd's vendoring is also optional to avoid potential import conflicts when embedding the etcd server or using the etcd client.
To build `etcd` from the `master` branch without a `GOPATH` using the official `build` script:
@ -27,7 +18,6 @@ To build `etcd` from the `master` branch without a `GOPATH` using the official `
$ git clone https://github.com/coreos/etcd.git
$ cd etcd
$ ./build
$ ./bin/etcd
```
To build a vendored `etcd` from the `master` branch via `go get`:
@ -37,7 +27,6 @@ To build a vendored `etcd` from the `master` branch via `go get`:
$ echo$GOPATH
/Users/example/go
$ go get github.com/coreos/etcd/cmd/etcd
$ $GOPATH/bin/etcd
```
To build `etcd` from the `master` branch without vendoring (may not build due to upstream conflicts):
@ -47,20 +36,28 @@ To build `etcd` from the `master` branch without vendoring (may not build due to
$ echo$GOPATH
/Users/example/go
$ go get github.com/coreos/etcd
$ $GOPATH/bin/etcd
```
## Test the installation
Check the etcd binary is built correctly by starting etcd and setting a key.
Start etcd:
### Starting etcd
If etcd is built without using GOPATH, run the following:
```
$ ./bin/etcd
```
If etcd is built using GOPATH, run the following:
Set a key:
```
$ $GOPATH/bin/etcd
```
### Setting a key
Run the following:
```
$ ETCDCTL_API=3 ./bin/etcdctl put foo bar
@ -73,4 +70,4 @@ If OK is printed, then etcd is working!
@ -12,58 +12,72 @@ The easiest way to get started using etcd as a distributed key-value store is to
- [Setting up local clusters][local_cluster]
- [Interacting with etcd][interacting]
-[API references][api_ref]
- [gRPC gateway][api_grpc_gateway]
-gRPC [etcd core][api_ref] and [etcd concurrency][api_concurrency_ref] API references
- [HTTP JSON API through the gRPC gateway][api_grpc_gateway]
- [gRPC naming and discovery][grpc_naming]
- [Client][namespace_client] and [proxy][namespace_proxy] namespacing
- [Embedding etcd][embed_etcd]
- [Experimental features and APIs][experimental]
- [System limits][system-limit]
## Operating etcd clusters
Administrators who need to create reliableand scalable key-value stores for the developers they support should begin with a [cluster on multiple machines][clustering].
Administrators who need a fault-tolerant etcd cluster for either development or production should begin with a [cluster on multiple machines][clustering].
- [Setting up etcd clusters][clustering]
- [Setting up etcd gateways][gateway]
- [Setting up etcd gRPC proxy (pre-alpha)][grpc_proxy]
#### Do clients have to send requests to the etcd leader?
### Do clients have to send requests to the etcd leader?
[Raft][raft] is leader-based; the leader handles all client requests which need cluster consensus. However, the client does not need to know which node is the leader. Any request that requires consensus sent to a follower is automatically forwarded to the leader. Requests that do not require consensus (e.g., serialized reads) can be processed by any cluster member.
### Configuration
## Configuration
#### What is the difference between advertise-urls and listen-urls?
### What is the difference between listen-<client,peer>-urls, advertise-client-urls or initial-advertise-peer-urls?
`listen-urls` specifies the local addresses etcd server binds to for accepting incoming connections. To listen on a port for all interfaces, specify `0.0.0.0` as the listen IP address.
`listen-client-urls` and `listen-peer-urls` specify the local addresses etcd server binds to for accepting incoming connections. To listen on a port for all interfaces, specify `0.0.0.0` as the listen IP address.
`advertise-urls` specifies the addresses etcd clients or other etcd members should use to contact the etcd server. The advertise addresses must be reachable from the remote machines. Do not advertise addresses like `localhost` or `0.0.0.0` for a production setup since these addresses are unreachable from remote machines.
`advertise-client-urls` and `initial-advertise-peer-urls` specify the addresses etcd clients or other etcd members should use to contact the etcd server. The advertise addresses must be reachable from the remote machines. Do not advertise addresses like `localhost` or `0.0.0.0` for a production setup since these addresses are unreachable from remote machines.
### Deployment
### Why doesn't changing `--listen-peer-urls` or `--initial-advertise-peer-urls` update the advertised peer URLs in `etcdctl member list`?
#### System requirements
A member's advertised peer URLs come from `--initial-advertise-peer-urls` on initial cluster boot. Changing the listen peer URLs or the initial advertise peers after booting the member won't affect the exported advertise peer URLs since changes must go through quorum to avoid membership configuration split brain. Use `etcdctl member update` to update a member's peer URLs.
## Deployment
### System requirements
Since etcd writes data to disk, SSD is highly recommended. To prevent performance degradation or unintentionally overloading the key-value store, etcd enforces a 2GB default storage size quota, configurable up to 8GB. To avoid swapping or running out of memory, the machine should have at least as much RAM to cover the quota. At CoreOS, an etcd cluster is usually deployed on dedicated CoreOS Container Linux machines with dual-core processors, 2GB of RAM, and 80GB of SSD *at the very least*. **Note that performance is intrinsically workload dependent; please test before production deployment**. See [hardware][hardware-setup] for more recommendations.
Most stable production environment is Linux operating system with amd64 architecture; see [supported platform][supported-platform] for more.
#### Why an odd number of cluster members?
### Why an odd number of cluster members?
An etcd cluster needs a majority of nodes, a quorum, to agree on updates to the cluster state. For a cluster with n members, quorum is (n/2)+1. For any odd-sized cluster, adding one node will always increase the number of nodes necessary for quorum. Although adding a node to an odd-sized cluster appears better since there are more machines, the fault tolerance is worse since exactly the same number of nodes may fail without losing quorum but there are more nodes that can fail. If the cluster is in a state where it can't tolerate any more failures, adding a node before removing nodes is dangerous because if the new node fails to register with the cluster (e.g., the address is misconfigured), quorum will be permanently lost.
#### What is maximum cluster size?
### What is maximum cluster size?
Theoretically, there is no hard limit. However, an etcd cluster probably should have no more than seven nodes. [Google Chubby lock service][chubby], similar to etcd and widely deployed within Google for many years, suggests running five nodes. A 5-member etcd cluster can tolerate two member failures, which is enough in most cases. Although larger clusters provide better fault tolerance, the write performance suffers because data must be replicated across more machines.
#### What is failure tolerance?
### What is failure tolerance?
An etcd cluster operates so long as a member quorum can be established. If quorum is lost through transient network failures (e.g., partitions), etcd automatically and safely resumes once the network recovers and restores quorum; Raft enforces cluster consistency. For power loss, etcd persists the Raft log to disk; etcd replays the log to the point of failure and resumes cluster participation. For permanent hardware failure, the node may be removed from the cluster through [runtime reconfiguration][runtime reconfiguration].
@ -50,19 +54,19 @@ It is recommended to have an odd number of members in a cluster. An odd-size clu
Adding a member to bring the size of cluster up to an even number doesn't buy additional fault tolerance. Likewise, during a network partition, an odd number of members guarantees that there will always be a majority partition that can continue to operate and be the source of truth when the partition ends.
#### Does etcd work in cross-region or cross data center deployments?
### Does etcd work in cross-region or cross data center deployments?
Deploying etcd across regions improves etcd's fault tolerance since members are in separate failure domains. The cost is higher consensus request latency from crossing data center boundaries. Since etcd relies on a member quorum for consensus, the latency from crossing data centers will be somewhat pronounced because at least a majority of cluster members must respond to consensus requests. Additionally, cluster data must be replicated across all peers, so there will be bandwidth cost as well.
With longer latencies, the default etcd configuration may cause frequent elections or heartbeat timeouts. See [tuning] for adjusting timeouts for high latency deployments.
### Operation
## Operation
#### How to backup a etcd cluster?
### How to backup a etcd cluster?
etcdctl provides a `snapshot` command to create backups. See [backup][backup] for more details.
#### Should I add a member before removing an unhealthy member?
### Should I add a member before removing an unhealthy member?
When replacing an etcd node, it's important to remove the member first and then add its replacement.
@ -74,20 +78,40 @@ Additionally, that new member is risky because it may turn out to be misconfigur
On the other hand, if the downed member is removed from cluster membership first, the number of members becomes 2 and the quorum remains at 2. Following that removal by adding a new member will also keep the quorum steady at 2. So, even if the new node can't be brought up, it's still possible to remove the new member through quorum on the remaining live members.
#### Why won't etcd accept my membership changes?
### Why won't etcd accept my membership changes?
etcd sets `strict-reconfig-check` in order to reject reconfiguration requests that would cause quorum loss. Abandoning quorum is really risky (especially when the cluster is already unhealthy). Although it may be tempting to disable quorum checking if there's quorum loss to add a new member, this could lead to full fledged cluster inconsistency. For many applications, this will make the problem even worse ("disk geometry corruption" being a candidate for most terrifying).
### Performance
### Why does etcd lose its leader from disk latency spikes?
#### How should I benchmark etcd?
This is intentional; disk latency is part of leader liveness. Suppose the cluster leader takes a minute to fsync a raft log update to disk, but the etcd cluster has a one second election timeout. Even though the leader can process network messages within the election interval (e.g., send heartbeats), it's effectively unavailable because it can't commit any new proposals; it's waiting on the slow disk. If the cluster frequently loses its leader due to disk latencies, try [tuning][tuning] the disk settings or etcd time parameters.
### What does the etcd warning "request ignored (cluster ID mismatch)" mean?
Every new etcd cluster generates a new cluster ID based on the initial cluster configuration and a user-provided unique `initial-cluster-token` value. By having unique cluster ID's, etcd is protected from cross-cluster interaction which could corrupt the cluster.
Usually this warning happens after tearing down an old cluster, then reusing some of the peer addresses for the new cluster. If any etcd process from the old cluster is still running it will try to contact the new cluster. The new cluster will recognize a cluster ID mismatch, then ignore the request and emit this warning. This warning is often cleared by ensuring peer addresses among distinct clusters are disjoint.
### What does "mvcc: database space exceeded" mean and how do I fix it?
The [multi-version concurrency control][api-mvcc] data model in etcd keeps an exact history of the keyspace. Without periodically compacting this history (e.g., by setting `--auto-compaction`), etcd will eventually exhaust its storage space. If etcd runs low on storage space, it raises a space quota alarm to protect the cluster from further writes. So long as the alarm is raised, etcd responds to write requests with the error `mvcc: database space exceeded`.
To recover from the low space quota alarm:
1. [Compact][maintenance-compact] etcd's history.
2. [Defragment][maintenance-defragment] every etcd endpoint.
3. [Disarm][maintenance-disarm] the alarm.
## Performance
### How should I benchmark etcd?
Try the [benchmark] tool. Current [benchmark results][benchmark-result] are available for comparison.
#### What does the etcd warning "apply entries took too long" mean?
### What does the etcd warning "apply entries took too long" mean?
After a majority of etcd members agree to commit a request, each etcd server applies the request to its data store and persists the result to disk. Even with a slow mechanical disk or a virtualized network disk, such as Amazon’s EBS or Google’s PD, applying a request should normally take fewer than 50 milliseconds. If the average apply duration exceeds 100 milliseconds, etcd will warn that entries are taking too long to apply.
Usually this issue is caused by a slow disk. The disk could be experiencing contention among etcd and other applications, or the disk is too simply slow (e.g., a shared virtualized disk). To rule out a slow disk from causing this warning, monitor [backend_commit_duration_seconds][backend_commit_metrics] (p99 duration should be less than 25ms) to confirm the disk is reasonably fast. If the disk is too slow, assigning a dedicated disk to etcd or using faster disk will typically solve the problem.
The second most common cause is CPU starvation. If monitoring of the machine’s CPU usage shows heavy utilization, there may not be enough compute capacity for etcd. Moving etcd to dedicated machine, increasing process resource isolation cgroups, or renicing the etcd server process into a higher priority can usually solve the problem.
@ -96,7 +120,7 @@ Expensive user requests which access too many keys (e.g., fetching the entire ke
If none of the above suggestions clear the warnings, please [open an issue][new_issue] with detailed logging, monitoring, metrics and optionally workload information.
#### What does the etcd warning "failed to send out heartbeat on time" mean?
### What does the etcd warning "failed to send out heartbeat on time" mean?
etcd uses a leader-based consensus protocol for consistent data replication and log execution. Cluster members elect a single leader, all other members become followers. The elected leader must periodically send heartbeats to its followers to maintain its leadership. Followers infer leader failure if no heartbeats are received within an election interval and trigger an election. If a leader doesn’t send its heartbeats in time but is still running, the election is spurious and likely caused by insufficient resources. To catch these soft failures, if the leader skips two heartbeat intervals, etcd will warn it failed to send a heartbeat on time.
@ -108,11 +132,10 @@ A slow network can also cause this issue. If network metrics among the etcd mach
If none of the above suggestions clear the warnings, please [open an issue][new_issue] with detailed logging, monitoring, metrics and optionally workload information.
#### What does the etcd warning "request ignored (cluster ID mismatch)" mean?
### What does the etcd warning "snapshotting is taking more than x seconds to finish ..." mean?
Every new etcd cluster generates a new cluster ID based on the initial cluster configuration and a user-provided unique `initial-cluster-token` value. By having unique cluster ID's, etcd is protected from cross-cluster interaction which could corrupt the cluster.
etcd sends a snapshot of its complete key-value store to refresh slow followers and for [backups][backup]. Slow snapshot transfer times increase MTTR; if the cluster is ingesting data with high throughput, slow followers may livelock by needing a new snapshot before finishing receiving a snapshot. To catch slow snapshot performance, etcd warns when sending a snapshot takes more than thirty seconds and exceeds the expected transfer time for a 1Gbps connection.
Usually this warning happens after tearing down an old cluster, then reusing some of the peer addresses for the new cluster. If any etcd process from the old cluster is still running it will try to contact the new cluster. The new cluster will recognize a cluster ID mismatch, then ignore the request and emit this warning. This warning is often cleared by ensuring peer addresses among distinct clusters are disjoint.
- [etcd/clientv3](https://github.com/coreos/etcd/blob/master/clientv3) - the officially maintained Go client for v3
- [etcd/client](https://github.com/coreos/etcd/blob/master/client) - the officially maintained Go client for v2
- [go-etcd](https://github.com/coreos/go-etcd) - the deprecated official client. May be useful for older (<2.0.0)versionsofetcd.
- [encWrapper](https://github.com/lumjjb/etcd/tree/enc_wrapper/clientwrap/encwrapper) - encWrapper is an encryption wrapper for the etcd client Keys API/KV.
**Java libraries**
@ -39,14 +40,18 @@
**Python libraries**
- [kragniz/python-etcd3](https://github.com/kragniz/python-etcd3) - Work in progress client for v3
- [kragniz/python-etcd3](https://github.com/kragniz/python-etcd3) - Client for v3
- [blox/blox](https://github.com/blox/blox) - a collection of open source projects for container management and orchestration with AWS ECS
- [calavera/active-proxy](https://github.com/calavera/active-proxy) - HTTP Proxy configured with etcd
- [chain/chain](https://github.com/chain/chain) - software designed to operate and connect to highly scalable permissioned blockchain networks
- [chain/chain](https://github.com/chain/chain) - software designed to operate and connect to highly scalable permissioned blockchain networks
- [derekchiang/etcdplus](https://github.com/derekchiang/etcdplus) - A set of distributed synchronization primitives built upon etcd
- [go-discover](https://github.com/flynn/go-discover) - service discovery in Go
- [gleicon/goreman](https://github.com/gleicon/goreman/tree/etcd) - Branch of the Go Foreman clone with etcd support
@ -135,7 +147,6 @@
- [mattn/etcdenv](https://github.com/mattn/etcdenv) - "env" shebang with etcd integration
- [kelseyhightower/confd](https://github.com/kelseyhightower/confd) - Manage local app config files using templates and data from etcd
- [configdb](https://git.autistici.org/ai/configdb/tree/master) - A REST relational abstraction on top of arbitrary database backends, aimed at storing configs and inventories.
- [scrz](https://github.com/scrz/scrz) - Container manager, stores configuration in etcd.
- [fleet](https://github.com/coreos/fleet) - Distributed init system
- [kubernetes/kubernetes](https://github.com/kubernetes/kubernetes) - Container cluster manager introduced by Google.
- [mailgun/vulcand](https://github.com/mailgun/vulcand) - HTTP proxy that uses etcd as a configuration backend.
@ -146,3 +157,7 @@
- [lytics/metafora](https://github.com/lytics/metafora) - Go distributed task library
- [ryandoyle/nss-etcd](https://github.com/ryandoyle/nss-etcd) - A GNU libc NSS module for resolving names from etcd.
- [Gru](https://github.com/dnaeon/gru) - Orchestration made easy with Go
- [Vitess](http://vitess.io/) - Vitess is a database clustering system for horizontal scaling of MySQL.
- [lclarkmichalek/etcdhcp](https://github.com/lclarkmichalek/etcdhcp) - DHCP server that uses etcd for persistence and coordination.
- [openstack/networking-vpp](https://github.com/openstack/networking-vpp) - A networking driver that programs the [FD.io VPP dataplane](https://wiki.fd.io/view/VPP) to provide [OpenStack](https://www.openstack.org/) cloud virtual networking
- [openstack](https://github.com/openstack/governance/blob/master/reference/base-services.rst) - OpenStack services can rely on etcd as a base service.
This document is meant to give an overview of the etcd3 API's central design. It is by no means all encompassing, but intended to focus on the basic ideas needed to understand etcd without the distraction of less common API calls. All etcd3 API's are defined in [gRPC services][grpc-service], which categorize remote procedure calls (RPCs) understood by the etcd server. A full listing of all etcd RPCs are documented in markdown in the [gRPC API listing][grpc-api].
## Response header
## gRPC Services
All Responses from etcd API have a [response header][response_header] attached. The response header includes the metadata of the response.
Every API request sent to an etcd server is a gRPC remote procedure call. RPCs in etcd3 are categorized based on functionality into services.
Services important for dealing with etcd's key space include:
* KV - Creates, updates, fetches, and deletes key-value pairs.
* Watch - Monitors changes to keys.
* Lease - Primitives for consuming client keep-alive messages.
Services which manage the cluster itself include:
* Auth - Role based authentication mechanism for authenticating users.
* Cluster - Provides membership information and configuration facilities.
* Maintenance - Takes recovery snapshots, defragments the store, and returns per-member status information.
### Requests and Responses
All RPCs in etcd3 follow the same format. Each RPC has a function `Name` which takes `NameRequest` as an argument and returns `NameResponse` as a response. For example, here is the `Range` RPC description:
```protobuf
serviceKV{
Range(RangeRequest)returns(RangeResponse)
...
}
```
### Response header
All Responses from etcd API have an attached response header which includes cluster metadata for the response:
```proto
messageResponseHeader{
@ -15,24 +40,26 @@ message ResponseHeader {
}
```
* Cluster_ID - the ID of the cluster that generates the response
* Member_ID - the ID of the member that generates the response
* Revision - the revision of the key-value store when the response is generated
* Raft_Term - the Raft term of the member when the response is generated
* Cluster_ID - the ID of the cluster generating the response.
* Member_ID - the ID of the member generating the response.
* Revision - the revision of the key-value store when generating the response.
* Raft_Term - the Raft term of the member when generating the response.
An application may read the Cluster_ID (Member_ID) field to ensure it is communicating with the intended cluster (member).
Applications can use the `Revision` to know the latest revision of the key-value store. This is especially useful when applications specify a historical revision to make time `travel query` and wishes to know the latest revision at the time of the request.
Applications can use the `Revision` to know the latest revision of the key-value store. This is especially useful when applications specify a historical revision to make time `travel query` and wish to know the latest revision at the time of the request.
Applications can use `Raft_Term` to detect when the cluster completes a new leader election.
## Key-Value API
Key-Value API is used to manipulate key-value pairs stored inside etcd. The key-value API is defined as a [gRPC service][kv-service]. The Key-Value pair is defined as structured data in [protobuf format][kv-proto].
The Key-Value API manipulates key-value pairs stored inside etcd. The majority of requests made to etcd are usually key-value requests.
### System primitives
### Key-Value pair
A key-value pair is the smallest unit that the key-value API can manipulate. Each key-value pair has a number of fields:
A key-value pair is the smallest unit that the key-value API can manipulate. Each key-value pair has a number of fields, defined in [protobuf format][kv-proto]:
```protobuf
messageKeyValue{
@ -52,6 +79,403 @@ message KeyValue {
* Mod_Revision - revision of the last modification on the key.
* Lease - the ID of the lease attached to the key. If lease is 0, then no lease is attached to the key.
In addition to just the key and value, etcd attaches additional revision metadata as part of the key message. This revision information orders keys by time of creation and modification, which is useful for managing concurrency for distributed synchronization. The etcd client's [distributed shared locks][locks] use the creation revision to wait for lock ownership. Similarly, the modification revision is used for detecting [software transactional memory][STM] read set conflicts and waiting on [leader election][elections] updates.
#### Revisions
etcd maintains a 64-bit cluster-wide counter, the store revision, that is incremented each time the key space is modified. The revision serves as a global logical clock, sequentially ordering all updates to the store. The change represented by a new revisions is incremental; the data associated with a revision is the data that changed the store. Internally, a new revision means writing the changes to the backend's B+tree, keyed by the incremented revision.
Revisions become more valuable when taking considering etcd3's [multi-version concurrency control][mvcc] backend. The MVCC model means that the key-value store can be viewed from past revisions since historical key revisions are retained. The retention policy for this history can be configured by cluster administrators for fine-grained storage management; usually etcd3 discards old revisions of keys on a timer. A typical etcd3 cluster retains superseded key data for hours. This also buys reliable handling for long client disconnection, not just transient network disruptions: watchers simply resume from the last observed historical revision. Similarly, to read from the store at a particular point-in-time, read requests can be tagged with a revision to return keys from a view of the key space at the point in time that revision was committed.
#### Key ranges
The etcd3 data model indexes all keys over a flat binary key space. This differs from other key-value store systems that use a hierarchical system of organizing keys into directories. Instead of listing keys by directory, keys are listed by key intervals `[a, b)`.
These intervals are often referred to as "ranges" in etcd3. Operations over ranges are more powerful than operations on directories. Like a hierarchical store, intervals support single key lookups via `[a, a+1)` (e.g., ['a', 'a\x00') looks up 'a') and directory lookups by encoding keys by directory depth. In addition to those operations, intervals can also encode prefixes; for example the interval `['a', 'b')` looks up all keys prefixed by the string 'a'.
By convention, ranges for a Request are denoted by the fields `key` and `range_end`. The `key` field is the first key of the range and should be non-empty. The `range_end` is the key following the last key of the range. If `range_end` is not given or empty, the range is defined to contain only the key argument. If `range_end` is `key` plus one (e.g., "aa"+1 == "ab", "a\xff"+1 == "b"), then the range represents all keys prefixed with key. If both `key` and `range_end` are '\0', then range represents all keys. If `range_end` is '\0', the range is all keys greater than or equal to the key argument.
### Range
Keys are fetched from the key-value store using the `Range` API call, which takes a `RangeRequest`:
```protobuf
messageRangeRequest{
enumSortOrder{
NONE=0;// default, no sorting
ASCEND=1;// lowest target value first
DESCEND=2;// highest target value first
}
enumSortTarget{
KEY=0;
VERSION=1;
CREATE=2;
MOD=3;
VALUE=4;
}
byteskey=1;
bytesrange_end=2;
int64limit=3;
int64revision=4;
SortOrdersort_order=5;
SortTargetsort_target=6;
boolserializable=7;
boolkeys_only=8;
boolcount_only=9;
int64min_mod_revision=10;
int64max_mod_revision=11;
int64min_create_revision=12;
int64max_create_revision=13;
}
```
* Key, Range_End - The key range to fetch.
* Limit - the maximum number of keys returned for the request. When limit is set to 0, it is treated as no limit.
* Revision - the point-in-time of the key-value store to use for the range. If revision is less or equal to zero, the range is over the latest key-value store If the revision is compacted, ErrCompacted is returned as a response.
* Sort_Order - the ordering for sorted requests.
* Sort_Target - the key-value field to sort.
* Serializable - sets the range request to use serializable member-local reads. By default, Range is linearizable; it reflects the current consensus of the cluster. For better performance and availability, in exchange for possible stale reads, a serializable range request is served locally without needing to reach consensus with other nodes in the cluster.
* Keys_Only - return only the keys and not the values.
* Count_Only - return only the count of the keys in the range.
* Min_Mod_Revision - the lower bound for key mod revisions; filters out lesser mod revisions.
* Max_Mod_Revision - the upper bound for key mod revisions; filters out greater mod revisions.
* Min_Create_Revision - the lower bound for key create revisions; filters out lesser create revisions.
* Max_Create_Revision - the upper bound for key create revisions; filters out greater create revisions.
The client receives a `RangeResponse` message from the `Range` call:
```protobuf
messageRangeResponse{
ResponseHeaderheader=1;
repeatedmvccpb.KeyValuekvs=2;
boolmore=3;
int64count=4;
}
```
* Kvs - the list of key-value pairs matched by the range request. When `Count_Only` is set, `Kvs` is empty.
* More - indicates if there are more keys to return in the requested range if `limit` is set.
* Count - the total number of keys satisfying the range request.
### Put
Keys are saved into the key-value store by issuing a `Put` call, which takes a `PutRequest`:
```protobuf
messagePutRequest{
byteskey=1;
bytesvalue=2;
int64lease=3;
boolprev_kv=4;
boolignore_value=5;
boolignore_lease=6;
}
```
* Key - the name of the key to put into the key-value store.
* Value - the value, in bytes, to associate with the key in the key-value store.
* Lease - the lease ID to associate with the key in the key-value store. A lease value of 0 indicates no lease.
* Prev_Kv - when set, responds with the key-value pair data before the update from this `Put` request.
* Ignore_Value - when set, update the key without changing its current value. Returns an error if the key does not exist.
* Ignore_Lease - when set, update the key without changing its current lease. Returns an error if the key does not exist.
The client receives a `PutResponse` message from the `Put` call:
```protobuf
messagePutResponse{
ResponseHeaderheader=1;
mvccpb.KeyValueprev_kv=2;
}
```
* Prev_Kv - the key-value pair overwritten by the `Put`, if `Prev_Kv` was set in the `PutRequest`.
### Delete Range
Ranges of keys are deleted using the `DeleteRange` call, which takes a `DeleteRangeRequest`:
```protobuf
messageDeleteRangeRequest{
byteskey=1;
bytesrange_end=2;
boolprev_kv=3;
}
```
* Key, Range_End - The key range to delete.
* Prev_Kv - when set, return the contents of the deleted key-value pairs.
The client receives a `DeleteRangeResponse` message from the `DeleteRange` call:
```protobuf
messageDeleteRangeResponse{
ResponseHeaderheader=1;
int64deleted=2;
repeatedmvccpb.KeyValueprev_kvs=3;
}
```
* Deleted - number of keys deleted.
* Prev_Kv - a list of all key-value pairs deleted by the DeleteRange operation.
### Transaction
A transaction is an atomic If/Then/Else construct over the key-value store. It provides a primitive for grouping requests together in atomic blocks (i.e., then/else) whose execution is guarded (i.e., if) based on the contents of the key-value store. Transactions can be used for protecting keys from unintended concurrent updates, building compare-and-swap operations, and developing higher-level concurrency control.
A transaction can atomically process multiple requests in a single request. For modifications to the key-value store, this means the store's revision is incremented only once for the transaction and all events generated by the transaction will have the same revision. However, modifications to the same key multiple times within a single transaction are forbidden.
All transactions are guarded by a conjunction of comparisons, similar to an "If" statement. Each comparison checks a single key in the store. It may check for the absence or presence of a value, compare with a given value, or check a key's revision or version. Two different comparisons may apply to the same or different keys. All comparisons are applied atomically; if all comparisons are true, the transaction is said to succeed and etcd applies the transaction's then / `success` request block, otherwise it is said to fail and applies the else / `failure` request block.
Each comparison is encoded as a `Compare` message:
```protobuf
messageCompare{
enumCompareResult{
EQUAL=0;
GREATER=1;
LESS=2;
NOT_EQUAL=3;
}
enumCompareTarget{
VERSION=0;
CREATE=1;
MOD=2;
VALUE=3;
}
CompareResultresult=1;
// target is the key-value field to inspect for the comparison.
CompareTargettarget=2;
// key is the subject key for the comparison operation.
byteskey=3;
oneoftarget_union{
int64version=4;
int64create_revision=5;
int64mod_revision=6;
bytesvalue=7;
}
}
```
* Result - the kind of logical comparison operation (e.g., equal, less than, etc).
* Target - the key-value field to be compared. Either the key's version, create revision, modification revision, or value.
* Key - the key for the comparison.
* Target_Union - the user-specified data for the comparison.
After processing the comparison block, the transaction applies a block of requests. A block is a list of `RequestOp` messages:
```protobuf
messageRequestOp{
// request is a union of request types accepted by a transaction.
oneofrequest{
RangeRequestrequest_range=1;
PutRequestrequest_put=2;
DeleteRangeRequestrequest_delete_range=3;
}
}
```
* Request_Range - a `RangeRequest`.
* Request_Put - a `PutRequest`. The keys must be unique. It may not share keys with any other Puts or Deletes.
* Request_Delete_Range - a `DeleteRangeRequest`. It may not share keys with any Puts or Deletes requests.
All together, a transaction is issued with a `Txn` API call, which takes a `TxnRequest`:
```protobuf
messageTxnRequest{
repeatedComparecompare=1;
repeatedRequestOpsuccess=2;
repeatedRequestOpfailure=3;
}
```
* Compare - A list of predicates representing a conjunction of terms for guarding the transaction.
* Success - A list of requests to process if all compare tests evaluate to true.
* Failure - A list of requests to process if any compare test evaluates to false.
The client receives a `TxnResponse` message from the `Txn` call:
```protobuf
messageTxnResponse{
ResponseHeaderheader=1;
boolsucceeded=2;
repeatedResponseOpresponses=3;
}
```
* Succeeded - Whether `Compare` evaluated to true or false.
* Responses - A list of responses corresponding to the results from applying the `Success` block if succeeded is true or the `Failure` if succeeded is false.
The `Responses` list corresponds to the results from the applied `RequestOp` list, with each response encoded as a `ResponseOp`:
```protobuf
messageResponseOp{
oneofresponse{
RangeResponseresponse_range=1;
PutResponseresponse_put=2;
DeleteRangeResponseresponse_delete_range=3;
}
}
```
## Watch API
The Watch API provides an event-based interface for asynchronously monitoring changes to keys. An etcd3 watch waits for changes to keys by continuously watching from a given revision, either current or historical, and streams key updates back to the client.
### Events
Every change to every key is represented with `Event` messages. An `Event` message provides both the update's data and the type of update:
```protobuf
messageEvent{
enumEventType{
PUT=0;
DELETE=1;
}
EventTypetype=1;
KeyValuekv=2;
KeyValueprev_kv=3;
}
```
* Type - The kind of event. A PUT type indicates new data has been stored to the key. A DELETE indicates the key was deleted.
* KV - The KeyValue associated with the event. A PUT event contains current kv pair. A PUT event with kv.Version=1 indicates the creation of a key. A DELETE event contains the deleted key with its modification revision set to the revision of deletion.
* Prev_KV - The key-value pair for the key from the revision immediately before the event. To save bandwidth, it is only filled out if the watch has explicitly enabled it.
### Watch streams
Watches are long-running requests and use gRPC streams to stream event data. A watch stream is bi-directional; the client writes to the stream to establish watches and reads to receive watch event. A single watch stream can multiplex many distinct watches by tagging events with per-watch identifiers. This multiplexing helps reducing the memory footprint and connection overhead on the core etcd cluster.
Watches make three guarantees about events:
* Ordered - events are ordered by revision; an event will never appear on a watch if it precedes an event in time that has already been posted.
* Reliable - a sequence of events will never drop any subsequence of events; if there are events ordered in time as a <b<c,thenifthewatchreceiveseventsaandc,itisguaranteedtoreceiveb.
etcd is a consistent and durable key value store with mini-transaction(TODO: link to txn doc when we have it) support. The key value store is exposed through the KV APIs. etcd tries to ensure the strongest consistency and durability guarantees for a distributed system. This specification enumerates the KV API guarantees made by etcd.
etcd is a consistent and durable key value store with [mini-transaction][txn] support. The key value store is exposed through the KV APIs. etcd tries to ensure the strongest consistency and durability guarantees for a distributed system. This specification enumerates the KV API guarantees made by etcd.
### APIs to consider
@ -21,7 +21,7 @@ An etcd operation is considered complete when it is committed through consensus,
#### Revision
An etcd operation that modifies the key value store is assigned with a single increasing revision. A transaction operation might modifies the key value store multiple times, but only one revision is assigned. The revision attribute of a key value pair that modified by the operation has the same value as the revision of the operation. The revision can be used as a logical clock for key value store. A key value pair that has a larger revision is modified after a key value pair with a smaller revision. Two key value pairs that have the same revision are modified by an operation "concurrently".
An etcd operation that modifies the key value store is assigned a single increasing revision. A transaction operation might modify the key value store multiple times, but only one revision is assigned. The revision attribute of a key value pair that was modified by the operation has the same value as the revision of the operation. The revision can be used as a logical clock for key value store. A key value pair that has a larger revision is modified after a key value pair with a smaller revision. Two key value pairs that have the same revision are modified by an operation "concurrently".
### Guarantees provided
@ -61,3 +61,4 @@ etcd ensures linearizability for all other operations by default. Linearizabilit
The v3 protocol uses gRPC as its transport instead of a RESTful interface like v2. This new protocol provides an opportunity to iterate on and improve the v2 design. For example, v3 auth has connection based authentication, rather than v2's slower per-request authentication. Additionally, v2 auth's semantics tend to be unwieldy in practice with respect to reasoning about consistency, which will be described in the next sections. For v3, there is a well-defined description and implementation of the authentication mechanism which fixes the deficiencies in the v2 auth system.
### Functionality requirements
* Per connection authentication, not per request
* User ID + password based authentication implemented for the gRPC API
* Authentication must be refreshed after auth policy changes
* Its functionality should be as simple and useful as v2
* v3 provides a flat key space, unlike the directory structure of v2. Permission checking will be provided as interval matching.
* It should have stronger consistency guarantees than v2 auth
### Main required changes
* A client must create a dedicated connection only for authentication before sending authenticated requests
* Add permission information (user ID and authorized revision) to the Raft commands (`etcdserverpb.InternalRaftRequest`)
* Every request is permission checked in the state machine layer, rather than API layer
### Permission metadata consistency
The metadata for auth should also be stored and managed in the storage controlled by etcd's Raft protocol like other data stored in etcd. It is required for not sacrificing availability and consistency of the entire etcd cluster. If reading or writing the metadata (e.g. permission information) needs an agreement of every node (more than quorum), single node failure can stop the entire cluster. Requiring all nodes to agree at once means that checking ordinary read/write requests cannot be completed if any cluster member is down, even if the cluster has an available quorum. This unanimous scheme ultimately degrades cluster availability; quorum based consensus from raft should suffice since agreement follows from consistent ordering.
The authentication mechanism in the etcd v2 protocol has a tricky part because the metadata consistency should work as in the above, but does not: each permission check is processed by the etcd member that receives the client request (etcdserver/api/v2http/client.go), including follower members. Therefore, it's possible the check may be based on stale metadata.
This staleness means that auth configuration cannot be reflected as soon as operators execute etcdctl. Therefore there is no way to know how long the stale metadata is active. Practically, the configuration change is reflected immediately after the command execution. However, in some cases of heavy load, the inconsistent state can be prolonged and it might result in counter-intuitive situations for users and developers. It requires a workaround like this: https://github.com/coreos/etcd/pull/4317#issuecomment-179037582
### Inconsistent permissions are unsafe for linearized requests
Inconsistent authentication state is most serious for writes. Even if an operator disables write on a user, if the write is only ordered with respect to the key value store but not the authentication system, it's possible the write will complete successfully. Without ordering on both the auth store and the key-value store, the system will be susceptible to stale permission attacks.
Therefore, the permission checking logic should be added to the state machine of etcd. Each state machine should check the requests based on its permission information in the apply phase (so the auth information must not be stale).
## Design and implementation
### Authentication
At first, a client must create a gRPC connection only to authenticate its user ID and password. An etcd server will respond with an authentication reply. The reponse will be an authentication token on success or an error on failure. The client can use its authentication token to present its credentials to etcd when making API requests.
The client connection used to request the authentication token is typically thrown away; it cannot carry the new token's credentials. This is because gRPC doesn't provide a way for adding per RPC credential after creation of the connection (calling `grpc.Dial()`). Therefore, a client cannot assign a token to its connection that is obtained through the connection. The client needs a new connection for using the token.
#### Notes on the implementation of `Authenticate()` RPC
`Authenticate()` RPC generates an authentication token based on a given user name and password. etcd saves and checks a configured password and a given password using Go's `bcrypt` package. By design, `bcrypt`'s password checking mechanism is computationally expensive, taking nearly 100ms on an ordinary x64 server. Therefore, performing this check in the state machine apply phase would cause performance trouble: the entire etcd cluster can only serve almost 10 `Authenticate()` requests per second.
For good performance, the v3 auth mechanism checks passwords in etcd's API layer, where it can be parallelized outside of raft. However, this can lead to potential time-of-check/time-of-use (TOCTOU) permission lapses:
1. client A sends a request `Authenticate()`
1. the API layer processes the password checking part of `Authenticate()`
1. another client B sends a request of `ChangePassword()` and the server completes it
1. the state machine layer processes the part of getting a revision number for the `Authenticate()` from A
1. the server returns a success to A
1. now A is authenticated on an obsolete password
For avoiding such a situation, the API layer performs *version number validation* based on the revision number of the auth store. During password checking, the API layer saves the revision number of auth store. After successful password checking, the API layer compares the saved revision number and the latest revision number. If the numbers differ, it means someone else updated the auth metadata. So it retries the checking. With this mechanism, the successful password checking based on the obsolete password can be avoided.
### Resolving a token in the API layer
After authenticating with `Authenticate()`, a client can create a gRPC connection as it would without auth. In addition to the existing initialization process, the client must associate the token with the newly created connection. `grpc.WithPerRPCCredentials()` provides the functionality for this purpose.
Every authenticated request from the client has a token. The token can be obtained with `grpc.metadata.FromIncomingContext()` in the server side. The server can obtain who is issuing the request and when the user was authorized. The information will be filled by the API layer in the header (`etcdserverpb.RequestHeader.Username` and `etcdserverpb.RequestHeader.AuthRevision`) of a raft log entry (`etcdserverpb.InternalRaftRequest`).
### Checking permission in the state machine
The auth info in `etcdserverpb.RequestHeader` is checked in the apply phase of the state machine. This step checks the user is granted permission to requested keys on the latest revision of auth store.
### Two types of tokens: simple and JWT
There are two kinds of token types: simple and JWT. The simple token isn't designed for production use cases. Its tokens aren't cryptographically signed and servers must statefully track token-user correspondence; it is meant for development testing. JWT tokens should be used for production deployments since it is cryptographically signed and verified. From the implementation perspective, JWT is stateless. Its token can include metadata including username and revision, so servers don't need to remember correspondence between tokens and the metadata.
## Notes on the difference between KVS models and file system models
etcd v3 is a KVS, not a file system. So the permissions can be granted to the users in form of an exact key name or a key range like `["start key", "end key")`. It means that granting a permission of a nonexistent key is possible. Users should care about unintended permission granting. In a case of file system like system (e.g. Chubby or ZooKeeper), an inode like data structure can include the permission information. So granting permission to a nonexist key won't be possible (except the case of sticky bits).
The etcd v3 model requires multiple lookup of the metadata unlike the file system like systems. The worst case lookup cost will be sum the user's total granted keys and intervals. The cost cannot be avoided because v3's flat key space is completely different from Unix's file system model (every inode includes permission metadata). Practically the cost won’t be a serious problem because the metadata is small enough to benefit from caching.
etcd is designed to reliably store infrequently updated data and provide reliable watch queries. etcd exposes previous versions of key-value pairs to support inexpensive snapshots and watch history events (“time travel queries”). A persistent, multi-version, concurrency-control data model is a good fit for these use cases.
etcd stores data in a multiversion [persistent][persistent-ds] key-value store. The persistent key-value store preserves the previous version of a key-value pair when its value is superseded with new data. The key-value store is effectively immutable; its operations do not update the structure in-place, but instead always generates a new updated structure. All past versions of keys are still accessible and watchable after modification. To prevent the data store from growing indefinitely over time from maintaining old versions, the store may be compacted to shed the oldest versions of superseded data.
etcd stores data in a multiversion [persistent][persistent-ds] key-value store. The persistent key-value store preserves the previous version of a key-value pair when its value is superseded with new data. The key-value store is effectively immutable; its operations do not update the structure in-place, but instead always generate a new updated structure. All past versions of keys are still accessible and watchable after modification. To prevent the data store from growing indefinitely over time and from maintaining old versions, the store may be compacted to shed the oldest versions of superseded data.
### Logical view
The store’s logical view is a flat binary key space. The key space has a lexically sorted index on byte string keys so range queries are inexpensive.
The key space maintains multiple revisions. Each atomic mutative operation (e.g., a transaction operation may contain multiple operations) creates a new revision on the key space. All data held by previous revisions remains unchanged. Old versions of key can still be accessed through previous revisions. Likewise, revisions are indexed as well; ranging over revisions with watchers is efficient. If the store is compacted to recover space, revisions before the compact revision will be removed.
The key space maintains multiple revisions. Each atomic mutative operation (e.g., a transaction operation may contain multiple operations) creates a new revision on the key space. All data held by previous revisions remains unchanged. Old versions of key can still be accessed through previous revisions. Likewise, revisions are indexed as well; ranging over revisions with watchers is efficient. If the store is compacted to save space, revisions before the compact revision will be removed.
A key’s lifetime spans a generation. Each key may have one or multiple generations. Creating a key increments the generation of that key, starting at 1 if the key never existed. Deleting a key generates a key tombstone, concluding the key’s current generation. Each modification of a key creates a new version of the key. Once a compaction happens, any generation ended before the given revision will be removed and values set before the compaction revision except the latest one will be removed.
A key’s lifetime spans a generation, denoted by its version. Each key may have one or multiple generations. Creating a key increments the version of that key, starting at 1 if the key never existed. Deleting a key generates a key tombstone, concluding the key’s current generation by resetting its version. Each modification of a key increments its version. Once a compaction happens, any version ended before the given revision will be removed and values set before the compaction revision except the latest one will be removed.
### Physical view
etcd stores the physical data as key-value pairs in a persistent [b+tree][b+tree]. Each revision of the store’s state only contains the delta from its previous revision to be efficient. A single revision may correspond to multiple keys in the tree.
etcd stores the physical data as key-value pairs in a persistent [b+tree][b+tree]. Each revision of the store’s state only contains the delta from its previous revision to be efficient. A single revision may correspond to multiple keys in the tree.
The key of key-value pair is a 3-tuple (major, sub, type). Major is the store revision holding the key. Sub differentiates among keys within the same revision. Type is an optional suffix for special value (e.g., `t` if the value contains a tombstone). The value of the key-value pair contains the modification from previous revision, thus one delta from previous revision. The b+tree is ordered by key in lexical byte-order. Ranged lookups over revision deltas are fast; this enables quickly finding modifications from one specific revision to another. Compaction removes out-of-date keys-value pairs.
The name "etcd" originated from two ideas, the unix "/etc" folder and "d"istibuted systems. The "/etc" folder is a place to store configuration data for a single system whereas etcd stores configuration information for large scale distributed systems. Hence, a "d"istributed "/etc" is "etcd".
etcd stores metadata in a consistent and fault-tolerant way. Distributed systems use etcd as a consistent key-value store for configuration management, service discovery, and coordinating distributed work. Common distributed patterns using etcd include leader election, [distributed locks][etcd-concurrency], and monitoring machine liveness.
etcd is designed as a general substrate for large scale distributed systems. These are systems that will never tolerate split-brain operation and are willing to sacrifice availability to achieve this end. etcd stores metadata in a consistent and fault-tolerant way. An etcd cluster is meant to provide key-value storage with best of class stability, reliability, scalability and performance.
Distributed systems use etcd as a consistent key-value store for configuration management, service discovery, and coordinating distributed work. Many [organizations][production-users] use etcd to implement production systems such as container schedulers, service discovery services, and distributed data storage. Common distributed patterns using etcd include [leader election][etcd-etcdctl-elect], [distributed locks][etcd-etcdctl-lock], and monitoring machine liveness.
## Use cases
- Container Linux by CoreOS: Application running on [Container Linux][container-linux] gets automatic, zero-downtime Linux kernel updates. Container Linux uses [locksmith] to coordinate updates. locksmith implements a distributed semaphore over etcd to ensure only a subset of a cluster is rebooting at any given time.
- Container Linux by CoreOS: Applications running on [Container Linux][container-linux] get automatic, zero-downtime Linux kernel updates. Container Linux uses [locksmith] to coordinate updates. Locksmith implements a distributed semaphore over etcd to ensure only a subset of a cluster is rebooting at any given time.
- [Kubernetes][kubernetes] stores configuration data into etcd for service discovery and cluster management; etcd's consistency is crucial for correctly scheduling and operating services. The Kubernetes API server persists cluster state into etcd. It uses etcd's watch API to monitor the cluster and roll out critical configuration changes.
## Comparison chart
## Features and system comparisons
Perhaps etcd already seems like a good fit, but as with all technological decisions, proceed with caution. Please note this documentation is written by the etcd team. Although the ideal is a disinterested comparison of technology and features, the authors’ expertise and biases obviously favor etcd. Use only as directed.
TODO
The table below is a handy quick reference for spotting the differences among etcd and its most popular alternatives at a glance. Further commentary and details for each column are in the sections following the table.
| Concurrency Primitives | [Lock RPCs][etcd-v3lock], [Election RPCs][etcd-v3election], [command line locks][etcd-etcdctl-lock], [command line elections][etcd-etcdctl-elect], [recipes][etcd-recipe] in go | External [curator recipes][curator] in Java | [Native lock API][consul-lock] | [Rare][newsql-leader], if any |
| Linearizable Reads | [Yes][etcd-linread] | No | [Yes][consul-linread] | Sometimes |
| Multi-version Concurrency Control | [Yes][etcd-mvcc] | No | No | Sometimes |
ZooKeeper solves the same problem as etcd: distributed system coordination and metadata storage. However, etcd has the luxury of hindsight taken from engineering and operational experience with ZooKeeper’s design and implementation. The lessons learned from Zookeeper certainly informed etcd’s design, helping it support large scale systems like Kubernetes. The improvements etcd made over Zookeeper include:
* Dynamic cluster membership reconfiguration
* Stable read/write under high load
* A multi-version concurrency control data model
* Reliable key monitoring which never silently drop events
* Lease primitives decoupling connections from sessions
* APIs for safe distributed shared locks
Furthermore, etcd supports a wide range of languages and frameworks out of the box. Whereas Zookeeper has its own custom Jute RPC protocol, which is totally unique to Zookeeper and limits its [supported language bindings][zk-bindings], etcd’s client protocol is built from [gRPC][grpc], a popular RPC framework with language bindings for go, C++, Java, and more. Likewise, gRPC can be serialized into JSON over HTTP, so even general command line utilities like `curl` can talk to it. Since systems can select from a variety of choices, they are built on etcd with native tooling rather than around etcd with a single fixed set of technologies.
When considering features, support, and stability, new applications planning to use Zookeeper for a consistent key value store would do well to choose etcd instead.
### Consul
Consul bills itself as an end-to-end service discovery framework. To wit, it includes services such as health checking, failure detection, and DNS. Incidentally, Consul also exposes a key value store with mediocre performance and an intricate API. As it stands in Consul 0.7, the storage system does not scales well; systems requiring millions of keys will suffer from high latencies and memory pressure. The key value API is missing, most notably, multi-version keys, conditional transactions, and reliable streaming watches.
etcd and Consul solve different problems. If looking for a distributed consistent key value store, etcd is a better choice over Consul. If looking for end-to-end cluster service discovery, etcd will not have enough features; choose Kubernetes, Consul, or SmartStack.
### NewSQL (Cloud Spanner, CockroachDB, TiDB)
Both etcd and NewSQL databases (e.g., [Cockroach][cockroach], [TiDB][tidb], [Google Spanner][spanner]) provide strong data consistency guarantees with high availability. However, the significantly different system design parameters lead to significantly different client APIs and performance characteristics.
NewSQL databases are meant to horizontally scale across data centers. These systems typically partition data across multiple consistent replication groups (shards), potentially distant, storing data sets on the order of terabytes and above. This sort of scaling makes them poor candidates for distributed coordination as they have long latencies from waiting on clocks and expect updates with mostly localized dependency graphs. The data is organized into tables, including SQL-style query facilities with richer semantics than etcd, but at the cost of additional complexity for processing, planning, and optimizing queries.
In short, choose etcd for storing metadata or coordinating distributed applications. If storing more than a few GB of data or if full SQL queries are needed, choose a NewSQL database.
## Using etcd for metadata
etcd replicates all data within a single consistent replication group. For storing up to a few GB of data with consistent ordering, this is the most efficient approach. Each modification of cluster state, which may change multiple keys, is assigned a global unique ID, called a revision in etcd, from a monotonically increasing counter for reasoning over ordering. Since there’s only a single replication group, the modification request only needs to go through the raft protocol to commit. By limiting consensus to one replication group, etcd gets distributed consistency with a simple protocol while achieving low latency and high throughput.
The replication behind etcd cannot horizontally scale because it lacks data sharding. In contrast, NewSQL databases usually shard data across multiple consistent replication groups, storing data sets on the order of terabytes and above. However, to assign each modification a global unique and increasing ID, each request must go through an additional coordination protocol among replication groups. This extra coordination step may potentially conflict on the global ID, forcing ordered requests to retry. The result is a more complicated approach with typically worse performance than etcd for strict ordering.
If an application reasons primarily about metadata or metadata ordering, such as to coordinate processes, choose etcd. If the application needs a large data store spanning multiple data centers and does not heavily depend on strong global ordering properties, choose a NewSQL database.
## Using etcd for distributed coordination
etcd has distributed coordination primitives such as event watches, leases, elections, and distributed shared locks out of the box. These primitives are both maintained and supported by the etcd developers; leaving these primitives to external libraries shirks the responsibility of developing foundational distributed software, essentially leaving the system incomplete. NewSQL databases usually expect these distributed coordination primitives to be authored by third parties. Likewise, ZooKeeper famously has a separate and independent [library][curator] of coordination recipes. Consul, which provides a native locking API, goes so far as to apologize that it’s “[not a bulletproof method][consul-bulletproof]”.
In theory, it’s possible to build these primitives atop any storage systems providing strong consistency. However, the algorithms tend to be subtle; it is easy to develop a locking algorithm that appears to work, only to suddenly break due to thundering herd and timing skew. Furthermore, other primitives supported by etcd, such as transactional memory depend on etcd’s MVCC data model; simple strong consistency is not enough.
For distributed coordination, choosing etcd can help prevent operational headaches and save engineering effort.
Authentication was added in etcd 2.1. The etcd v3 API slightly modified the authentication feature's API and user interface to better fit the new data model. This guide is intended to help users set up basic authentication and role-based access control in etcd v3.
## Special users and roles
There is one special user, `root`, and one special role, `root`.
### User `root`
The `root` user, which has full access to etcd, must be created before activating authentication. The idea behind the `root` user is for administrative purposes: managing roles and ordinary users. The `root` user must have the `root` role and is allowed to change anything inside etcd.
### Role `root`
The role `root` may be granted to any user, in addition to the root user. A user with the `root` role has both global read-write access and permission to update the cluster's authentication configuration. Furthermore, the `root` role grants privileges for general cluster maintenance, including modifying cluster membership, defragmenting the store, and taking snapshots.
## Working with users
The `user` subcommand for `etcdctl` handles all things having to do with user accounts.
A listing of users can be found with:
```
$ etcdctl user list
```
Creating a user is as easy as
```
$ etcdctl user add myusername
```
Creating a new user will prompt for a new password. The password can be supplied from standard input when an option `--interactive=false` is given.
Roles can be granted and revoked for a user with:
```
$ etcdctl user grant-role myusername foo
$ etcdctl user revoke-role myusername bar
```
The user's settings can be inspected with:
```
$ etcdctl user get myusername
```
And the password for a user can be changed with
```
$ etcdctl user passwd myusername
```
Changing the password will prompt again for a new password. The password can be supplied from standard input when an option `--interactive=false` is given.
Delete an account with:
```
$ etcdctl user delete myusername
```
## Working with roles
The `role` subcommand for `etcdctl` handles all things having to do with access controls for particular roles, as were granted to individual users.
List roles with:
```
$ etcdctl role list
```
Create a new role with:
```
$ etcdctl role add myrolename
```
A role has no password; it merely defines a new set of access rights.
Roles are granted access to a single key or a range of keys.
The range can be specified as an interval [start-key, end-key) where start-key should be lexically less than end-key in an alphabetical manner.
Access can be granted as either read, write, or both, as in the following examples:
```
# Give read access to a key /foo
$ etcdctl role grant-permission myrolename read /foo
# Give read access to keys with a prefix /foo/. The prefix is equal to the range [/foo/, /foo0)
$ etcdctl role grant-permission myrolename --prefix=true read /foo/
# Give write-only access to the key at /foo/bar
$ etcdctl role grant-permission myrolename write /foo/bar
# Give full access to keys in a range of [key1, key5)
$ etcdctl role grant-permission myrolename readwrite key1 key5
# Give full access to keys with a prefix /pub/
$ etcdctl role grant-permission myrolename --prefix=true readwrite /pub/
```
To see what's granted, we can look at the role at any time:
```
$ etcdctl role get myrolename
```
Revocation of permissions is done the same logical way:
```
$ etcdctl role revoke-permission myrolename /foo/bar
```
As is removing a role entirely:
```
$ etcdctl role remove myrolename
```
## Enabling authentication
The minimal steps to enabling auth are as follows. The administrator can set up users and roles before or after enabling authentication, as a matter of preference.
Make sure the root user is created:
```
$ etcdctl user add root
Password of root:
```
Enable authentication:
```
$ etcdctl auth enable
```
After this, etcd is running with authentication enabled. To disable it for any reason, use the reciprocal command:
```
$ etcdctl --user root:rootpw auth disable
```
## Using `etcdctl` to authenticate
`etcdctl` supports a similar flag as `curl` for authentication.
```
$ etcdctl --user user:password get foo
```
The password can be taken from a prompt:
```
$ etcdctl --user user get foo
```
Otherwise, all `etcdctl` commands remain the same. Users and roles can still be created and modified, but require authentication by a user with the root role.
## Using TLS Common Name
If an etcd server is launched with the option `--client-cert-auth=true`, the field of Common Name (CN) in the client's TLS cert will be used as an etcd user. In this case, the common name authenticates the user and the client does not need a password. Note that if both of 1. `--client-cert-auth=true` is passed and CN is provided by the client, and 2. username and password are provided by the client, the username and password based authentication is prioritized.
**Each member must have a different name flag specified or else discovery will fail due to duplicated names. `Hostname` or `machine-id` can be a good choice.**
**Each member must have a different name flag specified or else discovery will fail due to duplicated names. `Hostname` or `machine-id` can be a good choice.**
Now we start etcd with those relevant flags for each member:
@ -456,9 +456,11 @@ $ etcd --name infra2 \
--listen-peer-urls http://10.0.1.12:2380
```
Since v3.1.0 (except v3.2.9), when `etcd --discovery-srv=example.com` is configured with TLS, server will only authenticate peers/clients when the provided certs have root domain `example.com` as an entry in Subject Alternative Name (SAN) field. See [Notes for DNS SRV][security-guide-dns-srv].
### Gateway
etcd gateway is a simple TCP proxy that forwards network data to the etcd cluster. Please read [gateway guide] for more information.
etcd gateway is a simple TCP proxy that forwards network data to the etcd cluster. Please read [gateway guide][gateway] for more information.
### Proxy
@ -475,5 +477,6 @@ To setup an etcd cluster with proxies of v2 API, please read the the [clustering
@ -28,7 +28,7 @@ To start etcd automatically using custom settings at startup in Linux, using a [
### --snapshot-count
+ Number of committed transactions to trigger a snapshot to disk.
+ default: "10000"
+ default: "100000"
+ env variable: ETCD_SNAPSHOT_COUNT
### --heartbeat-interval
@ -69,9 +69,39 @@ To start etcd automatically using custom settings at startup in Linux, using a [
### --cors
+ Comma-separated white list of origins for CORS (cross-origin resource sharing).
+ default: none
+ default: ""
+ env variable: ETCD_CORS
### --quota-backend-bytes
+ Raise alarms when backend size exceeds the given quota (0 defaults to low space quota).
+ default: 0
+ env variable: ETCD_QUOTA_BACKEND_BYTES
### --max-txn-ops
+ Maximum number of operations permitted in a transaction.
+ default: 128
+ env variable: ETCD_MAX_TXN_OPS
### --max-request-bytes
+ Maximum client request size in bytes the server will accept.
+ default: 1572864
+ env variable: ETCD_MAX_REQUEST_BYTES
### --grpc-keepalive-min-time
+ Minimum duration interval that a client should wait before pinging server.
+ default: 5s
+ env variable: ETCD_GRPC_KEEPALIVE_MIN_TIME
### --grpc-keepalive-interval
+ Frequency duration of server-to-client ping to check if a connection is alive (0 to disable).
+ default: 2h
+ env variable: ETCD_GRPC_KEEPALIVE_INTERVAL
### --grpc-keepalive-timeout
+ Additional duration of wait before closing a non-responsive connection (0 to disable).
+ default: 20s
+ env variable: ETCD_GRPC_KEEPALIVE_TIMEOUT
## Clustering flags
`--initial` prefix flags are used in bootstrapping ([static bootstrap][build-cluster], [discovery-service bootstrap][discovery] or [runtime reconfiguration][reconfig]) a new member, and ignored when restarting an existing member.
@ -112,12 +142,12 @@ To start etcd automatically using custom settings at startup in Linux, using a [
### --discovery
+ Discovery URL used to bootstrap the cluster.
+ default: none
+ default: ""
+ env variable: ETCD_DISCOVERY
### --discovery-srv
+ DNS srv domain used to bootstrap the cluster.
+ default: none
+ default: ""
+ env variable: ETCD_DISCOVERY_SRV
### --discovery-fallback
@ -127,7 +157,7 @@ To start etcd automatically using custom settings at startup in Linux, using a [
### --discovery-proxy
+ HTTP proxy to use for traffic to discovery service.
+ default: none
+ default: ""
+ env variable: ETCD_DISCOVERY_PROXY
### --strict-reconfig-check
@ -140,6 +170,16 @@ To start etcd automatically using custom settings at startup in Linux, using a [
+ default: 0
+ env variable: ETCD_AUTO_COMPACTION_RETENTION
### --auto-compaction-mode
+ Interpret 'auto-compaction-retention' one of: periodic|revision. 'periodic' for duration based retention, defaulting to hours if no time unit is provided (e.g. '5m'). 'revision' for revision number based retention.
+ default: periodic
+ env variable: ETCD_AUTO_COMPACTION_MODE
### --enable-v2
+ Accept etcd V2 client requests
+ default: true
+ env variable: ETCD_ENABLE_V2
## Proxy flags
`--proxy` prefix flags configures etcd to run in [proxy mode][proxy]. "proxy" supports v2 API only.
@ -179,19 +219,22 @@ To start etcd automatically using custom settings at startup in Linux, using a [
The security flags help to [build a secure etcd cluster][security].
### --ca-file [DEPRECATED]
### --ca-file
**DEPRECATED**
+ Path to the client server TLS CA file. `--ca-file ca.crt` could be replaced by `--trusted-ca-file ca.crt --client-cert-auth` and etcd will perform the same.
+ default: none
+ default: ""
+ env variable: ETCD_CA_FILE
### --cert-file
+ Path to the client server TLS cert file.
+ default: none
+ default: ""
+ env variable: ETCD_CERT_FILE
### --key-file
+ Path to the client server TLS key file.
+ default: none
+ default: ""
+ env variable: ETCD_KEY_FILE
### --client-cert-auth
@ -199,9 +242,14 @@ The security flags help to [build a secure etcd cluster][security].
+ default: false
+ env variable: ETCD_CLIENT_CERT_AUTH
### --client-crl-file
+ Path to the client certificate revocation list file.
+ default: ""
+ env variable: ETCD_CLIENT_CRL_FILE
### --trusted-ca-file
+ Path to the client server TLS trusted CA key file.
+ default: none
+ Path to the client server TLS trusted CA cert file.
+ default: ""
+ env variable: ETCD_TRUSTED_CA_FILE
### --auto-tls
@ -209,19 +257,22 @@ The security flags help to [build a secure etcd cluster][security].
+ default: false
+ env variable: ETCD_AUTO_TLS
### --peer-ca-file [DEPRECATED]
### --peer-ca-file
**DEPRECATED**
+ Path to the peer server TLS CA file. `--peer-ca-file ca.crt` could be replaced by `--peer-trusted-ca-file ca.crt --peer-client-cert-auth` and etcd will perform the same.
+ default: none
+ default: ""
+ env variable: ETCD_PEER_CA_FILE
### --peer-cert-file
+ Path to the peer server TLS cert file.
+ default: none
+ default: ""
+ env variable: ETCD_PEER_CERT_FILE
### --peer-key-file
+ Path to the peer server TLS key file.
+ default: none
+ default: ""
+ env variable: ETCD_PEER_KEY_FILE
### --peer-client-cert-auth
@ -229,9 +280,14 @@ The security flags help to [build a secure etcd cluster][security].
+ default: false
+ env variable: ETCD_PEER_CLIENT_CERT_AUTH
### --peer-crl-file
+ Path to the peer certificate revocation list file.
+ default: ""
+ env variable: ETCD_PEER_CRL_FILE
### --peer-trusted-ca-file
+ Path to the peer server TLS trusted CA file.
+ default: none
+ default: ""
+ env variable: ETCD_PEER_TRUSTED_CA_FILE
### --peer-auto-tls
@ -239,6 +295,11 @@ The security flags help to [build a secure etcd cluster][security].
+ default: false
+ env variable: ETCD_PEER_AUTO_TLS
### --peer-cert-allowed-cn
+ Allowed CommonName for inter peer authentication.
+ default: none
+ env variable: ETCD_PEER_CERT_ALLOWED_CN
## Logging flags
### --debug
@ -248,10 +309,9 @@ The security flags help to [build a secure etcd cluster][security].
### --log-package-levels
+ Set individual etcd subpackages to specific log levels. An example being `etcdserver=WARNING,security=DEBUG`
+ default: none (INFO for all packages)
+ default: "" (INFO for all packages)
+ env variable: ETCD_LOG_PACKAGE_LEVELS
## Unsafe flags
Please be CAUTIOUS when using unsafe flags because it will break the guarantees given by the consensus protocol.
@ -271,7 +331,7 @@ Follow the instructions when using these flags.
### --config-file
+ Load server configuration from a file.
+ default: none
+ default: ""
## Profiling flags
@ -283,6 +343,23 @@ Follow the instructions when using these flags.
+ Set level of detail for exported metrics, specify 'extensive' to include histogram metrics.
+ default: basic
### --listen-metrics-urls
+ List of URLs to listen on for metrics.
+ default: ""
## Auth flags
### --auth-token
+ Specify a token type and token specific options, especially for JWT. Its format is "type,var1=val1,var2=val2,...". Possible type is 'simple' or 'jwt'. Possible variables are 'sign-method' for specifying a sign method of jwt (its possible values are 'ES256', 'ES384', 'ES512', 'HS256', 'HS384', 'HS512', 'RS256', 'RS384', 'RS512', 'PS256', 'PS384', or 'PS512'), 'pub-key' for specifying a path to a public key for verifying jwt, and 'priv-key' for specifying a path to a private key for signing jwt.
+ Example option of JWT: '--auth-token jwt,pub-key=app.rsa.pub,priv-key=app.rsa,sign-method=RS512'
+ default: "simple"
## Experimental flags
### --experimental-corrupt-check-time
+ Duration of time between cluster corruption check passes
@ -68,9 +68,55 @@ Production clusters which refer to peers by DNS name known to the local resolver
In order to expose the etcd API to clients outside of Docker host, use the host IP address of the container. Please see [`docker inspect`](https://docs.docker.com/engine/reference/commandline/inspect) for more detail on how to get the IP address. Alternatively, specify `--net=host` flag to `docker run` command to skip placing the container inside of a separate network stack.
To provision a 3 node etcd cluster on bare-metal, you might find the examples in the [baremetal repo](https://github.com/coreos/coreos-baremetal/tree/master/examples) useful.
To provision a 3 node etcd cluster on bare-metal, the examples in the [baremetal repo](https://github.com/coreos/coreos-baremetal/tree/master/examples) may be useful.
## Mounting a certificate volume
The etcd release container does not include default root certificates. To use HTTPS with certificates trusted by a root authority (e.g., for discovery), mount a certificate directory into the etcd container:
# alert if another failed member will result in an unavailable cluster
ALERT InsufficientMembers
IF count(up{job="etcd"} == 0) > (count(up{job="etcd"}) / 2 - 1)
FOR 3m
LABELS {
severity = "critical"
}
ANNOTATIONS {
summary = "etcd cluster insufficient members",
description = "If one more etcd member goes down the cluster will be unavailable",
}
# etcd leader alerts
# ==================
# alert if any etcd instance has no leader
ALERT NoLeader
IF etcd_server_has_leader{job="etcd"} == 0
FOR 1m
LABELS {
severity = "critical"
}
ANNOTATIONS {
summary = "etcd member has no leader",
description = "etcd member {{ $labels.instance }} has no leader",
}
# alert if there are lots of leader changes
ALERT HighNumberOfLeaderChanges
IF increase(etcd_server_leader_changes_seen_total{job="etcd"}[1h]) > 3
LABELS {
severity = "warning"
}
ANNOTATIONS {
summary = "a high number of leader changes within the etcd cluster are happening",
description = "etcd instance {{ $labels.instance }} has seen {{ $value }} leader changes within the last hour",
}
# gRPC request alerts
# ===================
# alert if more than 1% of gRPC method calls have failed within the last 5 minutes
ALERT HighNumberOfFailedGRPCRequests
IF sum by(grpc_method) (rate(etcd_grpc_requests_failed_total{job="etcd"}[5m]))
/ sum by(grpc_method) (rate(etcd_grpc_total{job="etcd"}[5m])) > 0.01
FOR 10m
LABELS {
severity = "warning"
}
ANNOTATIONS {
summary = "a high number of gRPC requests are failing",
description = "{{ $value }}% of requests for {{ $labels.grpc_method }} failed on etcd instance {{ $labels.instance }}",
}
# alert if more than 5% of gRPC method calls have failed within the last 5 minutes
ALERT HighNumberOfFailedGRPCRequests
IF sum by(grpc_method) (rate(etcd_grpc_requests_failed_total{job="etcd"}[5m]))
/ sum by(grpc_method) (rate(etcd_grpc_total{job="etcd"}[5m])) > 0.05
FOR 5m
LABELS {
severity = "critical"
}
ANNOTATIONS {
summary = "a high number of gRPC requests are failing",
description = "{{ $value }}% of requests for {{ $labels.grpc_method }} failed on etcd instance {{ $labels.instance }}",
}
# alert if the 99th percentile of gRPC method calls take more than 150ms
ALERT GRPCRequestsSlow
IF histogram_quantile(0.99, sum(rate(grpc_server_handling_seconds_bucket{job="etcd",grpc_type="unary"}[5m])) by (grpc_service, grpc_method, le)) > 0.15
FOR 10m
LABELS {
severity = "critical"
}
ANNOTATIONS {
summary = "slow gRPC requests",
description = "on etcd instance {{ $labels.instance }} gRPC requests to {{ $labels.grpc_method }} are slow",
}
# HTTP requests alerts
# ====================
# alert if more than 1% of requests to an HTTP endpoint have failed within the last 5 minutes
ALERT HighNumberOfFailedHTTPRequests
IF sum(rate(grpc_server_handled_total{grpc_code!="OK",job="etcd"}[5m])) BY (grpc_service, grpc_method)
/ sum(rate(grpc_server_handled_total{job="etcd"}[5m])) BY (grpc_service, grpc_method) > 0.01
FOR 10m
LABELS {
severity = "warning"
}
ANNOTATIONS {
summary = "a high number of HTTP requests are failing",
description = "{{ $value }}% of requests for {{ $labels.method }} failed on etcd instance {{ $labels.instance }}",
}
# alert if more than 5% of requests to an HTTP endpoint have failed within the last 5 minutes
ALERT HighNumberOfFailedHTTPRequests
IF sum(rate(grpc_server_handled_total{grpc_code!="OK",job="etcd"}[5m])) BY (grpc_service, grpc_method)
/ sum(rate(grpc_server_handled_total{job="etcd"}[5m])) BY (grpc_service, grpc_method) > 0.05
FOR 5m
LABELS {
severity = "critical"
}
ANNOTATIONS {
summary = "a high number of HTTP requests are failing",
description = "{{ $value }}% of requests for {{ $labels.method }} failed on etcd instance {{ $labels.instance }}",
}
# alert if the 99th percentile of HTTP requests take more than 150ms
ALERT HTTPRequestsSlow
IF histogram_quantile(0.99, rate(etcd_http_successful_duration_seconds_bucket[5m])) > 0.15
FOR 10m
LABELS {
severity = "warning"
}
ANNOTATIONS {
summary = "slow HTTP requests",
description = "on etcd instance {{ $labels.instance }} HTTP requests to {{ $labels.method }} are slow",
Failures are common in a large deployment of machines. A machine fails when its hardware or software malfunctions. Multiple machines fail together when there are power failures or network issues. Multiple kinds of failures can also happen at once; it is almost impossible to enumerate all possible failure cases.
etcd gateway is a simple TCP proxy that forwards network data to the etcd cluster. The gateway is stateless and transparent; it neither inspects client requests nor interferes with cluster responses.
The gateway supports multiple etcd server endpoints. When the gateway starts, it randomly picks one etcd server endpoint and forwards all requests to that endpoint. This endpoint serves all requests until the gateway detects a network failure. If the gateway detects an endpoint failure, it will switch to a different endpoint, if available, to hide failures from its clients. Other retry policies, such as weighted round-robin, may be supported in the future.
The gateway supports multiple etcd server endpoints and works on a simple round-robin policy. It only routes to available enpoints and hides failures from its clients. Other retry policies, such as weighted round-robin, may be supported in the future.
## When to use etcd gateway
Every application that accesses etcd must first have the address of an etcd cluster client endpoint. If multiple applications on the same server access the same etcd cluster, every application still needs to know the advertised client endpoints of the etcd cluster. If the etcd cluster is reconfigured to have different endpoints, every application may also need to update its endpoint list. This wide-scale reconfiguration is both tedious and error prone.
etcd gateway solves this problem by serving as a stable local endpoint. A typical etcd gateway configuration has
each machine running a gateway listening on a local address and every etcd application connecting to its local gateway. The upshot is only the gateway needs to update its endpoints instead of updating each and every application.
etcd gateway solves this problem by serving as a stable local endpoint. A typical etcd gateway configuration has each machine running a gateway listening on a local address and every etcd application connecting to its local gateway. The upshot is only the gateway needs to update its endpoints instead of updating each and every application.
In summary, to automatically propagate cluster endpoint changes, the etcd gateway runs on every machine serving multiple applications accessing same etcd cluster.
In summary, to automatically propagate cluster endpoint changes, the etcd gateway runs on every machine serving multiple applications accessing the same etcd cluster.
## When not to use etcd gateway
@ -63,4 +62,44 @@ Start the etcd gateway to fetch the endpoints from the DNS SRV entries with the
```bash
$ etcd gateway --discovery-srv=example.com
2016-08-16 11:21:18.867350 I | tcpproxy: ready to proxy client requests to [...]
```
```
## Configuration flags
### etcd cluster
#### --endpoints
* Comma-separated list of etcd server targets for forwarding client connections.
* Default: `127.0.0.1:2379`
* Invalid example: `https://127.0.0.1:2379` (gateway does not terminate TLS)
#### --discovery-srv
* DNS domain used to bootstrap cluster endpoints through SRV recrods.
* Default: (not set)
### Network
#### --listen-addr
* Interface and port to bind for accepting client requests.
* Default: `127.0.0.1:23790`
#### --retry-delay
* Duration of delay before retrying to connect to failed endpoints.
* Default: 1m0s
* Invalid example: "123" (expects time unit in format)
### Security
#### --insecure-discovery
* Accept SRV records that are insecure or susceptible to man-in-the-middle attacks.
* Default: `false`
#### --trusted-ca-file
* Path to the client TLS CA file for the etcd cluster. Used to authenticate endpoints.
*This is an alpha feature, we are looking for early feedback.*
The gRPC proxy is a stateless etcd reverse proxy operating at the gRPC layer (L7). The proxy is designed to reduce the total processing load on the core etcd cluster. For horizontal scalability, it coalesces watch and lease API requests. To protect the cluster against abusive clients, it caches key range requests.
The gRPC proxy supports multiple etcd server endpoints. When the proxy starts, it randomly picks one etcd server endpoint to use. This endpoint serves all requests until the proxy detects an endpoint failure. If the gRPC proxy detects an endpoint failure, it switches to a different endpoint, if available, to hide failures from its clients. Other retry policies, such as weighted round-robin, may be supported in the future.
@ -42,7 +40,30 @@ These two limitations should not cause problems for most use cases. In the futur
## Scalable lease API
TODO
To keep its leases alive, a client must establish at least one gRPC stream to an etcd server for sending periodic heartbeats. If an etcd workload involves heavy lease activity spread over many clients, these streams may contribute to excessive CPU utilization. To reduce the total number of streams on the core cluster, the proxy supports lease stream coalescing.
Assuming N clients are updating leases, a single gRPC proxy reduces the stream load on the etcd server from N to 1. Deployments may have additional gRPC proxies to further distribute streams across multiple proxies.
In the following example, three clients update three independent leases (`L1`, `L2`, and `L3`). The gRPC proxy coalesces the three client lease streams (`c-streams`) into a single lease keep alive stream (`s-stream`) attached to an etcd server. The proxy forwards client-side lease heartbeats from the c-streams to the s-stream, then returns the responses to the corresponding c-streams.
```
+-------------+
| etcd server |
+------+------+
^
| heartbeat L1, L2, L3
| (s-stream)
v
+-------+-----+
| gRPC proxy +<-----------+
+---+------+--+ | heartbeat L3
^ ^ | (c-stream)
heartbeat L1 | | heartbeat L2 |
(c-stream) v v (c-stream) v
+------+-+ +-+------+ +-----+--+
| client | | client | | client |
+--------+ +--------+ +--------+
```
## Abusive clients protection
@ -69,10 +90,136 @@ The etcd gRPC proxy starts and listens on port 8080. It forwards client requests
Sending requests through the proxy:
```bash
$ ETCDCTL_API=3./etcdctl --endpoints=127.0.0.1:2379 put foo bar
$ ETCDCTL_API=3 etcdctl --endpoints=127.0.0.1:2379 put foo bar
OK
$ ETCDCTL_API=3./etcdctl --endpoints=127.0.0.1:2379 get foo
$ ETCDCTL_API=3 etcdctl --endpoints=127.0.0.1:2379 get foo
foo
bar
```
## Client endpoint synchronization and name resolution
The proxy supports registering its endpoints for discovery by writing to a user-defined endpoint. This serves two purposes. First, it allows clients to synchronize their endpoints against a set of proxy endpoints for high availability. Second, it is an endpoint provider for etcd [gRPC naming](../dev-guide/grpc_naming.md).
Register proxy(s) by providing a user-defined prefix:
Suppose an application expects full control over the entire key space, but the etcd cluster is shared with other applications. To let all appications run without interfering with each other, the proxy can partition the etcd keyspace so clients appear to have access to the complete keyspace. When the proxy is given the flag `--namespace`, all client requests going into the proxy are translated to have a user-defined prefix on the keys. Accesses to the etcd cluster will be under the prefix and responses from the proxy will strip away the prefix; to the client, it appears as if there is no prefix at all.
To namespace a proxy, start it with `--namespace`:
The space quota in `etcd` ensures the cluster operates in a reliable fashion. Without a space quota, `etcd` may suffer from poor performance if the keyspace grows excessively large, or it may simply run out of storage space, leading to unpredictable cluster behavior. If the keyspace's backend database for any member exceeds the space quota, `etcd` raises a cluster-wide alarm that puts the cluster into a maintenance mode which only accepts key reads and deletes. Only after freeing enough space in the keyspace and defragmenting the backend database, along with clearing the space quota alarm can the cluster resume normal operation.
Each etcd server exports metrics under the `/metrics` path on its client port.
Each etcd server provides local monitoring information on its client port through http endpoints. The monitoring data is useful for both system health checking and cluster debugging.
## Debug endpoint
If `--debug` is set, the etcd server exports debugging information on its client port under the `/debug` path. Take care when setting `--debug`, since there will be degraded performance and verbose logging.
The `/debug/pprof` endpoint is the standard go runtime profiling endpoint. This can be used to profile CPU, heap, mutex, and goroutine utilization. For example, here `go tool pprof` gets the top 10 functions where etcd spends its time:
```sh
$ go tool pprof http://localhost:2379/debug/pprof/profile
Fetching profile from http://localhost:2379/debug/pprof/profile
Please wait... (30s)
Saved profile in /home/etcd/pprof/pprof.etcd.localhost:2379.samples.cpu.001.pb.gz
The `/debug/requests` endpoint gives gRPC traces and performance statistics through a web browser. For example, here is a `Range` request for the key `abc`:
tar -xvzf /tmp/prometheus-$PROMETHEUS_VERSION.linux-amd64.tar.gz --directory /tmp/ --strip-components=1
/tmp/prometheus -version
@ -56,7 +98,13 @@ nohup /tmp/prometheus \
Now Prometheus will scrape etcd metrics every 10 seconds.
## Grafana
### Alerting
There is a set of default alerts for etcd v3 clusters for [Prometheus 1.x](./etcd3_alert.rules) as well as [Prometheus 2.x](./etcd3_alert.rules.yml).
> Note: `job` labels may need to be adjusted to fit a particular need. The rules were written to apply to a single cluster so it is recommended to choose labels unique to a cluster.
### Grafana
[Grafana][grafana] has built-in Prometheus support; just add a Prometheus data source:
- etcd v3 master branch (commit SHA d8f325d), Go 1.6.2
- Ubuntu 17.04
- etcd 3.2.0, go 1.8.3
With this configuration, etcd can approximately write:
| Number of keys | Key size in bytes | Value size in bytes | Number of connections | Number of clients | Target etcd server | Average write QPS | Average latency per request | Memory |
| Number of keys | Key size in bytes | Value size in bytes | Number of connections | Number of clients | Target etcd server | Average write QPS | Average latency per request | Average server RSS |
put --key-size=8 --sequential-keys --total=100000 --val-size=256
```
Linearizable read requests go through a quorum of cluster members for consensus to fetch the most recent data. Serializable read requests are cheaper than linearizable reads since they are served by any single etcd member, instead of a quorum of members, in exchange for possibly serving stale data. etcd can read:
| Number of requests | Key size in bytes | Value size in bytes | Number of connections | Number of clients | Consistency | Average latency per request | Average read QPS |
| Number of requests | Key size in bytes | Value size in bytes | Number of connections | Number of clients | Consistency | Average read QPS | Average latency per request |
We encourage running the benchmark test when setting up an etcd cluster for the first time in a new environment to ensure the cluster achieves adequate performance; cluster latency and throughput can be sensitive to minor environment differences.
We encourage running the benchmark test when setting up an etcd cluster for the first time in a new environment to ensure the cluster achieves adequate performance; cluster latency and throughput can be sensitive to minor environment differences.
etcd is designed to withstand machine failures. An etcd cluster automatically recovers from temporary failures (e.g., machine reboots) and tolerates up to *(N-1)/2* permanent failures for a cluster of N members. When a member permanently fails, whether due to hardware failure or disk corruption, it loses access to the cluster. If the cluster permanently loses more than *(N-1)/2* members then it disastrously fails, irrevocably losing quorum. Once quorum is lost, the cluster cannot reach consensus and therefore cannot continue accepting updates.
@ -6,7 +6,7 @@ To recover from disastrous failure, etcd v3 provides snapshot and restore facili
Recovering a cluster first needs a snapshot of the keyspace from an etcd member. A snapshot may either be taken from a live member with the `etcdctl snapshot save` command or by copying the `member/snap/db` file from an etcd data directory. For example, the following command snapshots the keyspace served by `$ENDPOINT` to the file `snapshot.db`:
@ -14,7 +14,7 @@ Recovering a cluster first needs a snapshot of the keyspace from an etcd member.
$ ETCDCTL_API=3 etcdctl --endpoints $ENDPOINT snapshot save snapshot.db
```
### Restoring a cluster
## Restoring a cluster
To restore a cluster, all that is needed is a single snapshot "db" file. A cluster restore with `etcdctl snapshot restore` creates new etcd data directories; all members should restore using the same snapshot. Restoring overwrites some snapshot metadata (specifically, the member ID and cluster ID); the member loses its former identity. This metadata overwrite prevents the new member from inadvertently joining an existing cluster. Therefore in order to start a cluster from a snapshot, the restore must start a new logical cluster.
etcd comes with support for incremental runtime reconfiguration, which allows users to update the membership of the cluster at run time.
Reconfiguration requests can only be processed when the majority of the cluster members are functioning. It is **highly recommended** to always have a cluster size greater than two in production. It is unsafe to remove a member from a two member cluster. The majority of a two member cluster is also two. If there is a failure during the removal process, the cluster might not able to make progress and need to [restart from majority failure][majority failure].
Reconfiguration requests can only be processed when a majority of cluster members are functioning. It is **highly recommended** to always have a cluster size greater than two in production. It is unsafe to remove a member from a two member cluster. The majority of a two member cluster is also two. If there is a failure during the removal process, the cluster might not be able to make progress and need to [restart from majority failure][majority failure].
To better understand the design behind runtime reconfiguration, we suggest reading [the runtime reconfiguration document][runtime-reconf].
To better understand the design behind runtime reconfiguration, please read [the runtime reconfiguration document][runtime-reconf].
## Reconfiguration use cases
Let's walk through some common reasons for reconfiguring a cluster. Most of these just involve combinations of adding or removing a member, which are explained below under [Cluster Reconfiguration Operations][cluster-reconf].
This section will walk through some common reasons for reconfiguring a cluster. Most of these reasons just involve combinations of adding or removing a member, which are explained below under [Cluster Reconfiguration Operations][cluster-reconf].
### Cycle or upgrade multiple machines
If multiple cluster members need to move due to planned maintenance (hardware upgrades, network downtime, etc.), it is recommended to modify members one at a time.
It is safe to remove the leader, however there is a brief period of downtime while the election process takes place. If the cluster holds more than 50MB, it is recommended to [migrate the member's data directory][member migration].
It is safe to remove the leader, however there is a brief period of downtime while the election process takes place. If the cluster holds more than 50MB of v2 data, it is recommended to [migrate the member's data directory][member migration].
### Change the cluster size
Increasing the cluster size can enhance [failure tolerance][fault tolerance table] and provide better read performance. Since clients can read from any member, increasing the number of members increases the overall read throughput.
Increasing the cluster size can enhance [failure tolerance][fault tolerance table] and provide better read performance. Since clients can read from any member, increasing the number of members increases the overall serialized read throughput.
Decreasing the cluster size can improve the write performance of a cluster, with a trade-off of decreased resilience. Writes into the cluster are replicated to a majority of members of the cluster before considered committed. Decreasing the cluster size lowers the majority, and each write is committed more quickly.
@ -30,42 +30,34 @@ To replace the machine, follow the instructions for [removing the member][remove
### Restart cluster from majority failure
If the majority of the cluster is lost or all of the nodes have changed IP addresses, then manual action is necessary to recover safely.
The basic steps in the recovery process include [creating a new cluster using the old data][disaster recovery], forcing a single member to act as the leader, and finally using runtime configuration to [add new members][add member] to this new cluster one at a time.
If the majority of the cluster is lost or all of the nodes have changed IP addresses, then manual action is necessary to recover safely. The basic steps in the recovery process include [creating a new cluster using the old data][disaster recovery], forcing a single member to act as the leader, and finally using runtime configuration to [add new members][add member] to this new cluster one at a time.
## Cluster reconfiguration operations
Now that we have the use cases in mind, let us lay out the operations involved in each.
With these use cases in mind, the involved operations can be described for each.
Before making any change, the simple majority (quorum) of etcd members must be available.
This is essentially the same requirement as for any other write to etcd.
Before making any change, a simple majority (quorum) of etcd members must be available. This is essentially the same requirement for any kind of write to etcd.
All changes to the cluster are done one at a time:
All changes to the cluster must be done sequentially:
* To update a single member peerURLs, make an update operation
* To replace a single member, make an add then a remove operation
* To increase from 3 to 5 members, make two add operations
* To decrease from 5 to 3, make two remove operations
* To update a single member peerURLs, issue an update operation
* To replace a healthy single member, remove the old member then add a new member
* To increase from 3 to 5 members, issue two add operations
* To decrease from 5 to 3, issue two remove operations
All of these examples will use the `etcdctl` command line tool that ships with etcd.
To change membership without `etcdctl`, use the [v2 HTTP members API][member-api] or the [v3 gRPC members API][member-api-grpc].
All of these examples use the `etcdctl` command line tool that ships with etcd. To change membership without `etcdctl`, use the [v2 HTTP members API][member-api] or the [v3 gRPC members API][member-api-grpc].
### Update a member
#### Update advertise client URLs
To update the advertise client URLs of a member, simply restart
that member with updated client urls flag (`--advertise-client-urls`) or environment variable
(`ETCD_ADVERTISE_CLIENT_URLS`). The restarted member will self publish the updated URLs.
A wrongly updated client URL will not affect the health of the etcd cluster.
To update the advertise client URLs of a member, simply restart that member with updated client urls flag (`--advertise-client-urls`) or environment variable (`ETCD_ADVERTISE_CLIENT_URLS`). The restarted member will self publish the updated URLs. A wrongly updated client URL will not affect the health of the etcd cluster.
#### Update advertise peer URLs
To update the advertise peer URLs of a member, first update
it explicitly via member command and then restart the member. The additional action is required
since updating peer URLs changes the cluster wide configuration and can affect the health of the etcd cluster.
To update the advertise peer URLs of a member, first update it explicitly via member command and then restart the member. The additional action is required since updating peer URLs changes the cluster wide configuration and can affect the health of the etcd cluster.
To update the peer URLs, first, we need to find the target member's ID. To list all members with `etcdctl`:
To update the advertise peer URLs, first find the target member's ID. To list all members with `etcdctl`:
In this example let's`update` a8266ecf031671f3 member ID and change its peerURLs value to http://10.0.1.10:2380
This example will`update` a8266ecf031671f3 member ID and change its peerURLs value to `http://10.0.1.10:2380`:
```sh
$ etcdctl member update a8266ecf031671f3 http://10.0.1.10:2380
$ etcdctl member update a8266ecf031671f3 --peer-urls=http://10.0.1.10:2380
Updated member with ID a8266ecf031671f3 in cluster
```
### Remove a member
Let us say the member ID we want to remove is a8266ecf031671f3.
We then use the `remove` command to perform the removal:
Suppose the member ID to remove is a8266ecf031671f3. Use the `remove` command to perform the removal:
```sh
$ etcdctl member remove a8266ecf031671f3
@ -106,7 +97,7 @@ Adding a member is a two step process:
* Add the new member to the cluster via the [HTTP members API][member-api], the [gRPC members API][member-api-grpc], or the `etcdctl member add` command.
* Start the new member with the new cluster configuration, including a list of the updated members (existing members + the new member).
Using `etcdctl`let's add the new member to the cluster by specifying its [name][conf-name] and [advertised peer URLs][conf-adv-peer]:
`etcdctl`adds a new member to the cluster by specifying the member's [name][conf-name] and [advertised peer URLs][conf-adv-peer]:
`etcdctl` has informed the cluster about the new member and printed out the environment variables needed to successfully start it.
Now start the new etcd process with the relevant flags for the new member:
`etcdctl` has informed the cluster about the new member and printed out the environment variables needed to successfully start it. Now start the new etcd process with the relevant flags for the new member:
The new member will run as a part of the cluster and immediately begin catching up with the rest of the cluster.
If adding multiple members the best practice is to configure a single member at a time and verify it starts correctly before adding more new members.
If adding a new member to a 1-node cluster, the cluster cannot make progress before the new member starts because it needs two members as majority to agree on the consensus. This behavior only happens between the time `etcdctl member add` informs the cluster about the new member and the new member successfully establishing a connection to the existing one.
If adding multiple members the best practice is to configure a single member at a time and verify it starts correctly before adding more new members. If adding a new member to a 1-node cluster, the cluster cannot make progress before the new member starts because it needs two members as majority to agree on the consensus. This behavior only happens between the time `etcdctl member add` informs the cluster about the new member and the new member successfully establishing a connection to the existing one.
#### Error cases when adding members
In the following case we have not included our new host in the list of enumerated nodes.
If this is a new cluster, the node must be added to the list of initial cluster members.
In the following case a new host is not included in the list of enumerated nodes. If this is a new cluster, the node must be added to the list of initial cluster members.
```sh
$ etcd --name infra3 \
@ -145,7 +133,7 @@ etcdserver: assign ids error: the member count is unequal
exit1
```
In this case we give a different address (10.0.1.14:2380) to the one that we used to join the cluster (10.0.1.13:2380).
In this case, give a different address (10.0.1.14:2380) from the one used to join the cluster (10.0.1.13:2380):
```sh
$ etcd --name infra4 \
@ -155,7 +143,7 @@ etcdserver: assign ids error: unmatched member while checking PeerURLs
exit1
```
When we start etcd using the data directory of a removed member, etcd will exit automatically if it connects to any active member in the cluster:
If etcd starts using the data directory of a removed member, etcd automatically exits if it connects to any active member in the cluster:
@ -10,13 +10,13 @@ In etcd, every runtime reconfiguration has to go through [two phases][add-member
Phase 1 - Inform cluster of new configuration
To add a member into etcd cluster, make an API call to request a new member to be added to the cluster. This is only way to add a new member into an existing cluster. The API call returns when the cluster agrees on the configuration change.
To add a member into etcd cluster, make an API call to request a new member to be added to the cluster. This is the only way to add a new member into an existing cluster. The API call returns when the cluster agrees on the configuration change.
Phase 2 - Start new member
To join the etcd member into the existing cluster, specify the correct `initial-cluster` and set `initial-cluster-state` to `existing`. When the member starts, it will contact the existing cluster first and verify the current cluster configuration matches the expected one specified in `initial-cluster`. When the new member successfully starts, the cluster has reached the expected configuration.
To join the new etcd member into the existing cluster, specify the correct `initial-cluster` and set `initial-cluster-state` to `existing`. When the member starts, it will contact the existing cluster first and verify the current cluster configuration matches the expected one specified in `initial-cluster`. When the new member successfully starts, the cluster has reached the expected configuration.
By splitting the process into two discrete phases users are forced to be explicit regarding cluster membership changes. This actually gives users more flexibility and makes things easier to reason about. For example, if there is an attempt to add a new member with the same ID as an existing member in an etcd cluster, the action will fail immediately during phase one without impacting the running cluster. Similar protection is provided to prevent adding new members by mistake. If a new etcd member attempts to join the cluster before the cluster has accepted the configuration change,, it will not be accepted by the cluster.
By splitting the process into two discrete phases users are forced to be explicit regarding cluster membership changes. This actually gives users more flexibility and makes things easier to reason about. For example, if there is an attempt to add a new member with the same ID as an existing member in an etcd cluster, the action will fail immediately during phase one without impacting the running cluster. Similar protection is provided to prevent adding new members by mistake. If a new etcd member attempts to join the cluster before the cluster has accepted the configuration change, it will not be accepted by the cluster.
Without the explicit workflow around cluster membership etcd would be vulnerable to unexpected cluster membership changes. For example, if etcd is running under an init system such as systemd, etcd would be restarted after being removed via the membership API, and attempt to rejoin the cluster on startup. This cycle would continue every time a member is removed via the API and systemd is set to restart etcd after failing, which is unexpected.
@ -26,21 +26,21 @@ We expect runtime reconfiguration to be an infrequent operation. We decided to k
If a cluster permanently loses a majority of its members, a new cluster will need to be started from an old data directory to recover the previous state.
It is entirely possible to force removing the failed members from the existing cluster to recover. However, we decided not to support this method since it bypasses the normal consensus committing phase, which is unsafe. If the member to remove is not actually dead or force removed through different members in the same cluster, etcd will end up with a diverged cluster with same clusterID. This is very dangerous and hard to debug/fix afterwards.
It is entirely possible to force removing the failed members from the existing cluster to recover. However, we decided not to support this method since it bypasses the normal consensus committing phase, which is unsafe. If the member to remove is not actually dead or force removed through different members in the same cluster, etcd will end up with a diverged cluster with same clusterID. This is very dangerous and hard to debug/fix afterwards.
With a correct deployment, the possibility of permanent majority lose is very low. But it is a severe enough problem that worth special care. We strongly suggest reading the [disaster recovery documentation][disaster-recovery] and prepare for permanent majority lose before putting etcd into production.
With a correct deployment, the possibility of permanent majority lose is very low. But it is a severe enough problem that worth special care. We strongly suggest reading the [disaster recovery documentation][disaster-recovery] and preparing for permanent majority lose before putting etcd into production.
## Do not use public discovery service for runtime reconfiguration
The public discovery service should only be used for bootstrapping a cluster. To join member into an existing cluster, use runtime reconfiguration API.
The public discovery service should only be used for bootstrapping a cluster. To join member into an existing cluster, use runtime reconfiguration API.
Discovery service is designed for bootstrapping an etcd cluster in the cloud environment, when the IP addresses of all the members are not known beforehand. After successfully bootstrapping a cluster, the IP addresses of all the members are known. Technically, the discovery service should no longer be needed.
It seems that using public discovery service is a convenient way to do runtime reconfiguration, after all discovery service already has all the cluster configuration information. However relying on public discovery service brings troubles:
It seems that using public discovery service is a convenient way to do runtime reconfiguration, after all discovery service already has all the cluster configuration information. However relying on public discovery service brings troubles:
1. it introduces external dependencies for the entire life-cycle of the cluster, not just bootstrap time. If there is a network issue between the cluster and public discovery service, the cluster will suffer from it.
2. public discovery service must reflect correct runtime configuration of the cluster during it life-cycle. It has to provide security mechanism to avoid bad actions, and it is hard.
2. public discovery service must reflect correct runtime configuration of the cluster during it life-cycle. It has to provide security mechanism to avoid bad actions, and it is hard.
3. public discovery service has to keep tens of thousands of cluster configurations. Our public discovery service backend is not ready for that workload.
etcd supports automatic TLS as well as authentication through client certificates for both clients to server as well as peer (server to server / cluster) communication.
@ -16,7 +16,7 @@ etcd takes several certificate related configuration options, either through com
`--key-file=<path>`: Key for the certificate. Must be unencrypted.
`--client-cert-auth`: When this is set etcd will check all incoming HTTPS requests for a client certificate signed by the trusted CA, requests that don't supply a valid client certificate will fail.
`--client-cert-auth`: When this is set etcd will check all incoming HTTPS requests for a client certificate signed by the trusted CA, requests that don't supply a valid client certificate will fail. If [authentication][auth] is enabled, the certificate provides credentials for the user name given by the Common Name field.
Since v3.1.0 (except v3.2.9), discovery SRV bootstrapping authenticates `ServerName` with a root domain name from `--discovery-srv` flag. This is to avoid man-in-the-middle cert attacks, by requiring a certificate to have matching root domain name in its Subject Alternative Name (SAN) field. For instance, `etcd --discovery-srv=etcd.local` will only authenticate peers/clients when the provided certs have root domain `etcd.local` as an entry in Subject Alternative Name (SAN) field
## Notes for etcd proxy
etcd proxy terminates the TLS from its client if the connection is secure, and uses proxy's own key/cert specified in `--peer-key-file` and `--peer-cert-file` to communicate with etcd members.
@ -222,3 +226,4 @@ The certificate needs to be signed for the member's FQDN in its Subject Name, us
* etcd-maintainers are listed in https://github.com/coreos/etcd/blob/master/MAINTAINERS.
Experimental platforms appear to work in practice and have some platform specific code in etcd, but do not fully conform to the stable support policy. Unstable platforms have been lightly tested, but less than experimental. Unlisted architecture and operating system pairs are currently unsupported; caveat emptor.
### Supporting a new platform
## Supporting a new system platform
For etcd to officially support a new platform as stable, a few requirements are necessary to ensure acceptable quality:
@ -27,13 +28,13 @@ For etcd to officially support a new platform as stable, a few requirements are
4. Set up CI (TravisCI, SemaphoreCI or Jenkins) for running integration tests; etcd must pass intensive tests.
5. (Optional) Set up a functional testing cluster; an etcd cluster should survive stress testing.
### 32-bit and other unsupported systems
## 32-bit and other unsupported systems
etcd has known issues on 32-bit systems due to a bug in the Go runtime. See the [Go issue][go-issue] and [atomic package][go-atomic] for more information.
To avoid inadvertently running a possibly unstable etcd server, `etcd` on unstable or unsupported architectures will print a warning message and immediately exit if the environment variable `ETCD_UNSUPPORTED_ARCH` is not set to the target architecture.
Currently only the amd64 architecture is officially supported by `etcd`.
Currently amd64 and ppc64le architectures are officially supported by `etcd`.
@ -6,7 +6,7 @@ Migrating an application from the API v2 to the API v3 involves two steps: 1) mi
## Migrate client library
API v3 is different from API v2, thus application developers need to use a new client library to send requests to etcd API v3. The documentation of the client v3 is available at https://godoc.org/github.com/coreos/etcd/clientv3.
API v3 is different from API v2, thus application developers need to use a new client library to send requests to etcd API v3. The documentation of the client v3 is available at https://godoc.org/github.com/coreos/etcd/clientv3.
There are some notable differences between API v2 and API v3:
@ -22,6 +22,12 @@ There are some notable differences between API v2 and API v3:
Application data can be migrated either offline or online. Offline migration is much simpler than online migration and is recommended.
Sometimes an etcd cluster will possibly have v3 data which should not be overwritten. In this case, the migration process may want to confirm no v3 data is committed before proceeding. One way to check the cluster has no v3 keys is to issue the following `etcdctl` command, which scans the entire v3 keyspace for any key, expecting `0` as output:
```sh
ETCDCTL_API=3 etcdctl get "" --from-key --keys-only --limit 1| wc -l
```
### Offline migration
Offline migration is very simple but requires etcd downtime. If an etcd downtime window spanning from seconds to minutes is acceptable, offline migration is a good choice and is easy to automate.
@ -32,13 +38,17 @@ Second, migrate the v2 keys into v3 with the [migrate][migrate_command] (`ETCDCT
Restart the etcd members and everything should just work.
For etcd v3.3+, run `ETCDCTL_API=3 etcdctl endpoint hashkv --cluster` to ensure key-value stores are consistent post migration.
**Warn**: When v2 store has expiring TTL keys and migrate command intends to preserve TTLs, migration may be inconsistent with the last committed v2 state when run on any member with a raft index less than the last leader's raft index.
### Online migration
If the application cannot tolerate any downtime, then it must migrate online. The implementation of online migration will vary from application to application but the overall idea is the same.
First, write application code using the v3 API. The application must support two modes: a migration mode and a normal mode. The application starts in migration mode. When running in migration mode, the application reads keys using the v3 API first, and, if it cannot find the key, it retries with the API v2. In normal mode, the application only reads keys using the v3 API. The application writes keys over the API v3 in both modes. To acknowledge a switch from migration mode to normal mode, the application watches on a switch mode key. When switch key’s value turns to `true`, the application switches over from migration mode to normal mode.
Second, start a background job to migrate data from the store v2 to the mvcc store by reading keys from the API v2 and writing keys to the API v3.
Second, start a background job to migrate data from the store v2 to the mvcc store by reading keys from the API v2 and writing keys to the API v3.
After finishing data migration, the background job writes `true` into the switch mode key to notify the application that it may switch modes.
This guide assumes operational knowledge of Amazon Web Services (AWS), specifically Amazon Elastic Compute Cloud (EC2). This guide provides an introduction to design considerations when designing an etcd deployment on AWS EC2 and how AWS specific features may be utilized in that context.
## Capacity planning
As a critical building block for distributed systems it is crucial to perform adequate capacity planning in order to support the intended cluster workload. As a highly available and strongly consistent data store increasing the number of nodes in an etcd cluster will generally affect performance adversely. This makes sense intuitively, as more nodes means more members for the leader to coordinate state across. The most direct way to increase throughput and decrease latency of an etcd cluster is allocate more disk I/O, network I/O, CPU, and memory to cluster members. In the event it is impossible to temporarily divert incoming requests to the cluster, scaling the EC2 instances which comprise the etcd cluster members one at a time may improve performance. It is, however, best to avoid bottlenecks through capacity planning.
The etcd team has produced a [hardware recommendation guide](../op-guide/hardware.md) which is very useful for “ballparking” how many nodes and what instance type are necessary for a cluster.
AWS provides a service for creating groups of EC2 instances which are dynamically sized to match load on the instances. Using an Auto Scaling Group ([ASG](http://docs.aws.amazon.com/autoscaling/latest/userguide/AutoScalingGroup.html)) to dynamically scale an etcd cluster is not recommended for several reasons including:
* etcd performance is generally inversely proportional to the number of members in a cluster due to the synchronous replication which provides strong consistency of data stored in etcd
* the operational complexity of adding [lifecycle hooks](http://docs.aws.amazon.com/autoscaling/latest/userguide/lifecycle-hooks.html) to properly add and remove members from an etcd cluster by modifying the [runtime configuration](../op-guide/runtime-configuration.md)
Auto Scaling Groups do provide a number of benefits besides cluster scaling which include:
* distribution of EC2 instances across Availability Zones (AZs)
* EC2 instance fail over across AZs
* consolidated monitoring and life cycle control of instances within an ASG
The use of an ASG to create a [self healing etcd cluster](#self-healing) is one of the design considerations when deploying an etcd cluster to AWS.
## Cluster design
The purpose of this section is to provide foundational guidance for deploying etcd on AWS. The discussion will be framed by the following three critical design criteria about the etcd cluster itself:
* block device provider: limited to the tradeoffs between EBS or instance storage (InstanceStore)
* cluster topology: how many nodes should make up an etcd cluster; should these nodes be distributed over multiple AZs
* managing etcd members: creating a static cluster of EC2 instances or using an ASG.
The intended cluster workload should dictate the cluster design. A configuration store for microservices may require different design considerations than a distributed lock service, a secrets store, or a Kubernetes control plane. Cluster design tradeoffs include considerations such as:
* availability
* data durability after member failure
* performance/throughput
* self healing
### Availability
Instance availability on AWS is ultimately determined by the Amazon EC2 Region Service Level Agreement ([SLA](https://aws.amazon.com/ec2/sla/)) which is the policy by which Amazon describes their precise definition of a regional outage.
In the context of an etcd cluster this means a cluster must contain a minimum of three members where EC2 instances are spread across at least two AZs in order for an etcd cluster to be considered highly available at a Regional level.
For most use cases the additional latency associated with a cluster spanning across Availability Zones will introduce a negligible performance impact.
Availability considerations apply to all components of an application; if the application which accesses the etcd cluster will only be deployed to a single Availability Zone it may not make sense to make the etcd cluster highly available across zones.
### Data durability after member failure
A highly available etcd cluster is resilient to member loss, however, it is important to consider data durability in the event of disaster when designing an etcd deployment. Deploying etcd on AWS supports multiple mechanisms for data durability.
* replication: etcd replicates all data to all members of the etcd cluster. Therefore, given more members in the cluster and more independent failure domains, the less likely that data stored in an etcd cluster will be permanently lost in the event of disaster.
* Point in time etcd snapshotting: the etcd v3 API introduced support for snapshotting clusters. The operation is cheap enough (completing in the order of minutes) to run quite frequently and the resulting archives can be archived in a storage service like Amazon Simple Storage Service (S3).
* Amazon Elastic Block Storage (EBS): an EBS volume is a replicated network attached block device which have stronger storage safety guarantees than InstanceStore which has a life cycle associated with the life cycle of the attached EC2 instance. The life cycle of an EBS volume is not necessarily tied to an EC2 instance and can be detached and snapshotted independently which means that a single node etcd cluster backed by an EBS volume can provide a fairly reasonable level of data durability.
### Performance/Throughput
The performance of an etcd cluster is roughly quantifiable through latency and throughput metrics which are primarily affected by disk and network performance. Detailed performance planning information is provided in the [performance section](../op-guide/performance.md) of the etcd operations guide.
#### Network
AWS offers EC2 Placement Groups which allow the collocation of EC2 instances within a single Availability Zone which can be utilized in order to minimize network latency between etcd members in the cluster. It is important to remember that collocation of etcd nodes within a single AZ will provide weaker fault tolerance than distributing members across multiple AZs. [Enhanced networking for EC2 instances](http://docs.aws.amazon.com/AWSEC2/latest/UserGuide/enhanced-networking.html) may also improve network performance of individual EC2 instances.
#### Disk
AWS provides two basic types of block storage: [EBS volumes](https://aws.amazon.com/ebs/) and [EC2 Instance Store](http://docs.aws.amazon.com/AWSEC2/latest/UserGuide/InstanceStorage.html). As mentioned, an EBS volume is a network attached block device while instance storage is directly attached to the hypervisor of the EC2 host. EBS volumes will generally have higher latency, lower throughput, and greater performance variance than Instance Store volumes. If performance, rather than data safety, is the primary concern it is highly recommended that instance storage on the EC2 instances be utilized. Remember that the amount of available instance storage varies by EC2 [instance types](https://aws.amazon.com/ec2/instance-types/) which may impose additional performance considerations.
Inconsistent EBS volume performance can introduce etcd cluster instability. [Provisioned IOPS](http://docs.aws.amazon.com/AWSEC2/latest/UserGuide/EBSVolumeTypes.html#EBSVolumeTypes_piops) can provide more consistent performance than general purpose SSD EBS volumes. More information about EBS volume performance is available [from AWS](https://aws.amazon.com/ebs/details/) and Datadog has shared their experience with [getting optimal performance with AWS EBS Provisioned IOPS](https://www.datadoghq.com/blog/aws-ebs-provisioned-iops-getting-optimal-performance/) in their engineering blog.
### Self healing
While using an ASG to scale the size of an etcd cluster is not recommended, an ASG can be used effectively to maintain the desired number of nodes in the event of node failure. The maintenance of a stable number of etcd nodes will provide the etcd cluster with a measure of self healing.
### Next steps
The operational life cycle of an etcd cluster can be greatly simplified through the use of the etcd-operator. The open source etcd operator is a Kubernetes control plane operator which deploys and manages etcd clusters atop Kubernetes. While still in its early stages the etcd-operator already offers periodic backups to S3, detection and replacement of failed nodes, and automated disaster recovery from backups in the event of permanent quorum loss.
The following guide shows how to run etcd with [systemd][systemd-docs] under [Container Linux][container-linux-docs].
## Provisioning an etcd cluster
Cluster bootstrapping in Container Linux is simplest with [Ignition][container-linux-ignition]; `coreos-metadata.service` dynamically fetches the machine's IP for discovery. Note that etcd's discovery service protocol is only meant for bootstrapping, and cannot be used with runtime reconfiguration or cluster monitoring.
The [Container Linux Config Transpiler][container-linux-ct] compiles etcd configuration files into Ignition configuration files:
See [Container Linux Provisioning][container-linux-provision] for more details.
## etcd 3.x service
[Container Linux][container-linux-docs] does not include etcd 3.x binaries by default. Different versions of etcd 3.x can be fetched via `etcd-member.service`.
Confirm unit file exists:
```
systemctl cat etcd-member.service
```
Check if the etcd service is running:
```
systemctl status etcd-member.service
```
Example systemd drop-in unit to override the default service settings:
To see all runtime drop-in changes for system units:
```
systemd-delta --type=extended
```
To enable and start:
```
systemctl daemon-reload
systemctl enable --now etcd-member.service
```
To see the logs:
```
journalctl --unit etcd-member.service --lines 10
```
To stop and disable the service:
```
systemctl disable --now etcd-member.service
```
## etcd 2.x service
[Container Linux][container-linux-docs] includes a unit file `etcd2.service` for etcd 2.x, which will be removed in the near future. See [Container Linux FAQ][container-linux-faq] for more details.
Startingwithversion0.1.2bothetcdandetcdctlhavebeenportedtoFreeBSDandcan beinstalledeitherviapackagesorportssystem.Theirversionshavebeenrecently updatedto0.2.0sonowetcd and etcdctl can be enjoyedonFreeBSD10.0(RC4as ofnow)and9.x,wheretheyhavebeentested.Theymightalsoworkwheninstalledfrom portsonearlierversionsofFreeBSD,butit is untested; caveat emptor.
reproduceit on any otherplatform,like OSX or Linux),pleasesent a
problemreportusing thispage for more
information:http://www.freebsd.org/sendpr.html
Ifthere are anyissueswiththebuild/installprocedureorthere's aproblemthat islocaltoFreeBSDonly(forexample,bynotbeingableto reproduceit on any otherplatform,like OSX or Linux),pleasesend a problemreportusing thispage for more information:http://www.freebsd.org/sendpr.html
This document tracks people and use cases for etcd in production. By creating a list of production use cases we hope to build a community of advisors that we can reach out to with experience using various etcd applications, operation environments, and cluster sizes. The etcd development team may reach out periodically to check-in on how etcd is working in the field and update this list.
## All Kubernetes Users
- *Application*: https://kubernetes.io/
- *Environments*: AWS, OpenStack, Azure, Google Cloud, Huawei Cloud, Bare Metal, etc
**This is a meta user; please feel free to document specific Kubernetes clusters!**
All Kubernetes clusters use etcd as their primary data store. This means etcd's users include such companies as [Niantic, Inc Pokemon Go](https://cloudplatform.googleblog.com/2016/09/bringing-Pokemon-GO-to-life-on-Google-Cloud.html), [Box](https://blog.box.com/blog/kubernetes-box-microservices-maximum-velocity/), [CoreOS](https://coreos.com/tectonic), [Ticketmaster](https://www.youtube.com/watch?v=wqXVKneP0Hg), [Salesforce](https://www.salesforce.com) and many many more.
@ -50,7 +59,7 @@ Radius Intelligence uses Kubernetes running CoreOS to containerize and scale int
## Vonage
- *Application*: system configuration for microservices, scheduling, locks (future - service discovery)
- *Application*: kubernetes, vault backend, system configuration for microservices, scheduling, locks (future - service discovery)
- *Launched*: August 2015
- *Cluster Size*: 2 clusters of 5 members in 2 DCs, n local proxies 1-to-1 with microservice, (ssl and SRV look up)
- *Order of Data Size*: kilobytes
@ -58,6 +67,28 @@ Radius Intelligence uses Kubernetes running CoreOS to containerize and scale int
- *Environment*: VMWare, AWS
- *Backups*: Daily snapshots on VMs. Backups done for upgrades.
## PD
- *Application*: embed etcd
- *Launched*: Mar 2016
- *Cluster Size*: 3 or 5 members
- *Order of Data Size*: megabytes
- *Operator*: PingCAP, Inc.
- *Environment*: Bare Metal, AWS, etc.
- *Backups*: None.
PD(Placement Driver) is the central controller in the TiDB cluster. It saves the cluster meta information, schedule the data, allocate the global unique timestamp for the distributed transaction, etc. It embeds etcd to supply high availability and auto failover.
## Huawei
- *Application*: System configuration for overlay network (Canal)
@ -6,33 +6,16 @@ The network isn't the only source of latency. Each request and response may be i
## Time parameters
The underlying distributed consensus protocol relies on two separate time parameters to ensure that nodes can handoff leadership if one stalls or goes offline.
The first parameter is called the *Heartbeat Interval*.
This is the frequency with which the leader will notify followers that it is still the leader.
For best practices, the parameter should be set around round-trip time between members.
By default, etcd uses a `100ms` heartbeat interval.
The underlying distributed consensus protocol relies on two separate time parameters to ensure that nodes can handoff leadership if one stalls or goes offline. The first parameter is called the *Heartbeat Interval*. This is the frequency with which the leader will notify followers that it is still the leader.
For best practices, the parameter should be set around round-trip time between members. By default, etcd uses a `100ms` heartbeat interval.
The second parameter is the *Election Timeout*.
This timeout is how long a follower node will go without hearing a heartbeat before attempting to become leader itself.
By default, etcd uses a `1000ms` election timeout.
The second parameter is the *Election Timeout*. This timeout is how long a follower node will go without hearing a heartbeat before attempting to become leader itself. By default, etcd uses a `1000ms` election timeout.
Adjusting these values is a trade off.
The value of heartbeat interval is recommended to be around the maximum of average round-trip time (RTT) between members, normally around 0.5-1.5x the round-trip time.
If heartbeat interval is too low, etcd will send unnecessary messages that increase the usage of CPU and network resources.
On the other side, a too high heartbeat interval leads to high election timeout. Higher election timeout takes longer time to detect a leader failure.
The easiest way to measure round-trip time (RTT) is to use [PING utility][ping].
Adjusting these values is a trade off. The value of heartbeat interval is recommended to be around the maximum of average round-trip time (RTT) between members, normally around 0.5-1.5x the round-trip time. If heartbeat interval is too low, etcd will send unnecessary messages that increase the usage of CPU and network resources. On the other side, a too high heartbeat interval leads to high election timeout. Higher election timeout takes longer time to detect a leader failure. The easiest way to measure round-trip time (RTT) is to use [PING utility][ping].
The election timeout should be set based on the heartbeat interval and average round-trip time between members.
Election timeouts must be at least 10 times the round-trip time so it can account for variance in the network.
For example, if the round-trip time between members is 10ms then the election timeout should be at least 100ms.
The election timeout should be set based on the heartbeat interval and average round-trip time between members. Election timeouts must be at least 10 times the round-trip time so it can account for variance in the network. For example, if the round-trip time between members is 10ms then the election timeout should be at least 100ms.
The election timeout should be set to at least 5 to 10 times the heartbeat interval to account for variance in leader replication.
For a heartbeat interval of 50ms, set the election timeout to at least 250ms - 500ms.
The upper limit of election timeout is 50000ms (50s), which should only be used when deploying a globally-distributed etcd cluster.
A reasonable round-trip time for the continental United States is 130ms, and the time between US and Japan is around 350-400ms.
If the network has uneven performance or regular packet delays/loss then it is possible that a couple of retries may be necessary to successfully send a packet. So 5s is a safe upper limit of global round-trip time.
As the election timeout should be an order of magnitude bigger than broadcast time, in the case of ~5s for a globally distributed cluster, then 50 seconds becomes a reasonable maximum.
The upper limit of election timeout is 50000ms (50s), which should only be used when deploying a globally-distributed etcd cluster. A reasonable round-trip time for the continental United States is 130ms, and the time between US and Japan is around 350-400ms. If the network has uneven performance or regular packet delays/loss then it is possible that a couple of retries may be necessary to successfully send a packet. So 5s is a safe upper limit of global round-trip time. As the election timeout should be an order of magnitude bigger than broadcast time, in the case of ~5s for a globally distributed cluster, then 50 seconds becomes a reasonable maximum.
The heartbeat interval and election timeout value should be the same for all members in one cluster. Setting different values for etcd members may disrupt cluster stability.
@ -50,18 +33,13 @@ The values are specified in milliseconds.
## Snapshots
etcd appends all key changes to a log file.
This log grows forever and is a complete linear history of every change made to the keys.
A complete history works well for lightly used clusters but clusters that are heavily used would carry around a large log.
etcd appends all key changes to a log file. This log grows forever and is a complete linear history of every change made to the keys. A complete history works well for lightly used clusters but clusters that are heavily used would carry around a large log.
To avoid having a huge log etcd makes periodic snapshots.
These snapshots provide a way for etcd to compact the log by saving the current state of the system and removing old logs.
To avoid having a huge log etcd makes periodic snapshots. These snapshots provide a way for etcd to compact the log by saving the current state of the system and removing old logs.
### Snapshot tuning
Creating snapshots can be expensive so they're only created after a given number of changes to etcd.
By default, snapshots will be made after every 10,000 changes.
If etcd's memory usage and disk usage are too high, try lowering the snapshot threshold by setting the following on the command line:
Creating snapshots with the V2 backend can be expensive, so snapshots are only created after a given number of changes to etcd. By default, snapshots will be made after every 10,000 changes. If etcd's memory usage and disk usage are too high, try lowering the snapshot threshold by setting the following on the command line:
```sh
# Command line arguments:
@ -71,6 +49,17 @@ $ etcd --snapshot-count=5000
$ ETCD_SNAPSHOT_COUNT=5000 etcd
```
## Disk
An etcd cluster is very sensitive to disk latencies. Since etcd must persist proposals to its log, disk activity from other processes may cause long `fsync` latencies. The upshot is etcd may miss heartbeats, causing request timeouts and temporary leader loss. An etcd server can sometimes stably run alongside these processes when given a high disk priority.
On Linux, etcd's disk priority can be configured with `ionice`:
```sh
# best effort, highest priority
$ sudo ionice -c2 -n0 -p `pgrep etcd`
```
## Network
If the etcd leader serves a large number of concurrent client requests, it may delay processing follower peer requests due to network congestion. This manifests as send buffer error messages on the follower nodes:
@ -10,7 +10,7 @@ Before [starting an upgrade](#upgrade-procedure), read through the rest of this
#### Upgrade requirements
To upgrade an existing etcd deployment to 3.0, the running cluster must be 2.3 or greater. If it's before 2.3, please upgrade to [2.3](https://github.com/coreos/etcd/releases/tag/v2.3.0) before upgrading to 3.0.
To upgrade an existing etcd deployment to 3.0, the running cluster must be 2.3 or greater. If it's before 2.3, please upgrade to [2.3](https://github.com/coreos/etcd/releases/tag/v2.3.8) before upgrading to 3.0.
Also, to ensure a smooth rolling upgrade, the running cluster must be healthy. Check the health of the cluster by using the `etcdctl cluster-health` command before proceeding.
@ -52,7 +52,7 @@ member 8211f1d0f64f3269 is healthy: got healthy result from http://localhost:123
cluster is healthy
$ curl http://localhost:2379/version
{"etcdserver":"2.3.x","etcdcluster":"2.3.0"}
{"etcdserver":"2.3.x","etcdcluster":"2.3.8"}
```
#### 2. Stop the existing etcd process
@ -116,4 +116,14 @@ $ ETCDCTL_API=3 etcdctl endpoint health
127.0.0.1:22379 is healthy: successfully committed proposal: took = 18.513301ms
```
## Further considerations
- etcdctl environment variables have been updated. If `ETCDCTL_API=2 etcdctl cluster-health` works properly but `ETCDCTL_API=3 etcdctl endpoints health` responds with `Error: grpc: timed out when dialing`, be sure to use the [new variable names](https://github.com/coreos/etcd/tree/master/etcdctl#etcdctl).
## Known Issues
- etcd < v3.1 does not work properly if built with Go > v1.7. See [Issue 6951](https://github.com/coreos/etcd/issues/6951) for additional information.
- If an error such as `transport: http2Client.notifyError got notified that the client transport was broken unexpected EOF.` shows up in the etcd server logs, be sure etcd is a pre-built release or built with (etcd v3.1+ & go v1.7+) or (etcd <v3.1 & go v1.6.x).
- Adding a v3 node to v2.3 cluster during upgrades is not supported and could trigger panics. See [Issue 7249](https://github.com/coreos/etcd/issues/7429) for additional information. Mixed versions of etcd members are only allowed during v3 migration. Finish upgrades before making any membership changes.
@ -8,9 +8,18 @@ Before [starting an upgrade](#upgrade-procedure), read through the rest of this
### Upgrade checklists
#### Monitoring
Following metrics from v3.0.x have been deprecated in favor of [go-grpc-prometheus](https://github.com/grpc-ecosystem/go-grpc-prometheus):
-`etcd_grpc_requests_total`
-`etcd_grpc_requests_failed_total`
-`etcd_grpc_active_streams`
-`etcd_grpc_unary_requests_duration_seconds`
#### Upgrade requirements
To upgrade an existing etcd deployment to 3.1, the running cluster must be 3.0 or greater. If it's before 3.0, please upgrade to [3.0](https://github.com/coreos/etcd/releases/tag/v3.0.16) before upgrading to 3.1.
To upgrade an existing etcd deployment to 3.1, the running cluster must be 3.0 or greater. If it's before 3.0, please [upgrade to 3.0](upgrade_3_0.md) before upgrading to 3.1.
Also, to ensure a smooth rolling upgrade, the running cluster must be healthy. Check the health of the cluster by using the `etcdctl endpoint health` command before proceeding.
In the general case, upgrading from etcd 3.1 to 3.2 can be a zero-downtime, rolling upgrade:
- one by one, stop the etcd v3.1 processes and replace them with etcd v3.2 processes
- after running all v3.2 processes, new features in v3.2 are available to the cluster
Before [starting an upgrade](#upgrade-procedure), read through the rest of this guide to prepare.
### Client upgrade checklists
3.2 introduces two breaking changes.
Previously, `clientv3.Lease.TimeToLive` API returned `lease.ErrLeaseNotFound` on non-existent lease ID. 3.2 instead returns TTL=-1 in its response and no error (see [#7305](https://github.com/coreos/etcd/pull/7305)).
Before
```go
// when leaseID does not exist
resp,err:=TimeToLive(ctx,leaseID)
resp==nil
err==lease.ErrLeaseNotFound
```
After
```go
// when leaseID does not exist
resp,err:=TimeToLive(ctx,leaseID)
resp.TTL==-1
err==nil
```
`clientv3.NewFromConfigFile` is moved to `yaml.NewConfig`.
>=3.2.10 upgrades `grpc/grpc-go` to >=v1.7.x from v1.2.1, which introduces some breaking changes.
Previously, `grpc.ErrClientConnTimeout` error is returned on client dial time-outs. >=3.2.10 instead returns `context.DeadlineExceeded` (see [#8504](https://github.com/coreos/etcd/issues/8504)).
Before
```go
// expect dial time-out on ipv4 blackhole
_,err:=clientv3.New(clientv3.Config{
Endpoints:[]string{"http://254.0.0.1:12345"},
DialTimeout:2*time.Second
})
iferr==grpc.ErrClientConnTimeout{
// handle errors
}
```
After
```go
_,err:=clientv3.New(clientv3.Config{
Endpoints:[]string{"http://254.0.0.1:12345"},
DialTimeout:2*time.Second
})
iferr==context.DeadlineExceeded{
// handle errors
}
```
### Server upgrade checklists
#### Upgrade requirements
To upgrade an existing etcd deployment to 3.2, the running cluster must be 3.1 or greater. If it's before 3.1, please [upgrade to 3.1](upgrade_3_1.md) before upgrading to 3.2.
Also, to ensure a smooth rolling upgrade, the running cluster must be healthy. Check the health of the cluster by using the `etcdctl endpoint health` command before proceeding.
#### Preparation
Before upgrading etcd, always test the services relying on etcd in a staging environment before deploying the upgrade to the production environment.
Before beginning, [backup the etcd data](../op-guide/maintenance.md#snapshot-backup). Should something go wrong with the upgrade, it is possible to use this backup to [downgrade](#downgrade) back to existing etcd version. Please note that the `snapshot` command only backs up the v3 data. For v2 data, see [backing up v2 datastore](../v2/admin_guide.md#backing-up-the-datastore).
#### Mixed versions
While upgrading, an etcd cluster supports mixed versions of etcd members, and operates with the protocol of the lowest common version. The cluster is only considered upgraded once all of its members are upgraded to version 3.2. Internally, etcd members negotiate with each other to determine the overall cluster version, which controls the reported version and the supported features.
#### Limitations
Note: If the cluster only has v3 data and no v2 data, it is not subject to this limitation.
If the cluster is serving a v2 data set larger than 50MB, each newly upgraded member may take up to two minutes to catch up with the existing cluster. Check the size of a recent snapshot to estimate the total data size. In other words, it is safest to wait for 2 minutes between upgrading each member.
For a much larger total data size, 100MB or more , this one-time process might take even more time. Administrators of very large etcd clusters of this magnitude can feel free to contact the [etcd team][etcd-contact] before upgrading, and we'll be happy to provide advice on the procedure.
#### Downgrade
If all members have been upgraded to v3.2, the cluster will be upgraded to v3.2, and downgrade from this completed state is **not possible**. If any single member is still v3.1, however, the cluster and its operations remains "v3.1", and it is possible from this mixed cluster state to return to using a v3.1 etcd binary on all members.
Please [backup the data directory](../op-guide/maintenance.md#snapshot-backup) of all etcd members to make downgrading the cluster possible even after it has been completely upgraded.
### Upgrade procedure
This example shows how to upgrade a 3-member v3.1 ectd cluster running on a local machine.
#### 1. Check upgrade requirements
Is the cluster healthy and running v3.1.x?
```
$ ETCDCTL_API=3 etcdctl endpoint health --endpoints=localhost:2379,localhost:22379,localhost:32379
localhost:2379 is healthy: successfully committed proposal: took = 6.600684ms
localhost:22379 is healthy: successfully committed proposal: took = 8.540064ms
localhost:32379 is healthy: successfully committed proposal: took = 8.763432ms
$ curl http://localhost:2379/version
{"etcdserver":"3.1.7","etcdcluster":"3.1.0"}
```
#### 2. Stop the existing etcd process
When each etcd process is stopped, expected errors will be logged by other cluster members. This is normal since a cluster member connection has been (temporarily) broken:
```
2017-04-27 14:13:31.491746 I | raft: c89feb932daef420 [term 3] received MsgTimeoutNow from 6d4f535bae3ab960 and starts an election to get leadership.
2017-04-27 14:13:31.491769 I | raft: c89feb932daef420 became candidate at term 4
2017-04-27 14:13:31.491788 I | raft: c89feb932daef420 received MsgVoteResp from c89feb932daef420 at term 4
2017-04-27 14:13:31.491797 I | raft: c89feb932daef420 [logterm: 3, index: 9] sent MsgVote request to 6d4f535bae3ab960 at term 4
2017-04-27 14:13:31.491805 I | raft: c89feb932daef420 [logterm: 3, index: 9] sent MsgVote request to 9eda174c7df8a033 at term 4
2017-04-27 14:13:31.491815 I | raft: raft.node: c89feb932daef420 lost leader 6d4f535bae3ab960 at term 4
2017-04-27 14:13:31.524084 I | raft: c89feb932daef420 received MsgVoteResp from 6d4f535bae3ab960 at term 4
2017-04-27 14:13:31.524108 I | raft: c89feb932daef420 [quorum:2] has received 2 MsgVoteResp votes and 0 vote rejections
2017-04-27 14:13:31.524123 I | raft: c89feb932daef420 became leader at term 4
2017-04-27 14:13:31.524136 I | raft: raft.node: c89feb932daef420 elected leader c89feb932daef420 at term 4
2017-04-27 14:13:31.592650 W | rafthttp: lost the TCP streaming connection with peer 6d4f535bae3ab960 (stream MsgApp v2 reader)
2017-04-27 14:13:31.592825 W | rafthttp: lost the TCP streaming connection with peer 6d4f535bae3ab960 (stream Message reader)
2017-04-27 14:13:31.693275 E | rafthttp: failed to dial 6d4f535bae3ab960 on stream Message (dial tcp [::1]:2380: getsockopt: connection refused)
2017-04-27 14:13:31.693289 I | rafthttp: peer 6d4f535bae3ab960 became inactive
2017-04-27 14:13:31.936678 W | rafthttp: lost the TCP streaming connection with peer 6d4f535bae3ab960 (stream Message writer)
```
It's a good idea at this point to [backup the etcd data](../op-guide/maintenance.md#snapshot-backup) to provide a downgrade path should any problems occur:
```
$ etcdctl snapshot save backup.db
```
#### 3. Drop-in etcd v3.2 binary and start the new etcd process
The new v3.2 etcd will publish its information to the cluster:
```
2017-04-27 14:14:25.363225 I | etcdserver: published {Name:s1 ClientURLs:[http://localhost:2379]} to cluster a9ededbffcb1b1f1
```
Verify that each member, and then the entire cluster, becomes healthy with the new v3.2 etcd binary:
```
$ ETCDCTL_API=3 /etcdctl endpoint health --endpoints=localhost:2379,localhost:22379,localhost:32379
localhost:22379 is healthy: successfully committed proposal: took = 5.540129ms
localhost:32379 is healthy: successfully committed proposal: took = 7.321771ms
localhost:2379 is healthy: successfully committed proposal: took = 10.629901ms
```
Upgraded members will log warnings like the following until the entire cluster is upgraded. This is expected and will cease after all etcd cluster members are upgraded to v3.2:
```
2017-04-27 14:15:17.071804 W | etcdserver: member c89feb932daef420 has a higher version 3.2.0
2017-04-27 14:15:21.073110 W | etcdserver: the local etcd version 3.1.7 is not up-to-date
2017-04-27 14:15:21.073142 W | etcdserver: member 6d4f535bae3ab960 has a higher version 3.2.0
2017-04-27 14:15:21.073157 W | etcdserver: the local etcd version 3.1.7 is not up-to-date
2017-04-27 14:15:21.073164 W | etcdserver: member c89feb932daef420 has a higher version 3.2.0
```
#### 4. Repeat step 2 to step 3 for all other members
#### 5. Finish
When all members are upgraded, the cluster will report upgrading to 3.2 successfully:
```
2017-04-27 14:15:54.536901 N | etcdserver/membership: updated the cluster version from 3.1 to 3.2
2017-04-27 14:15:54.537035 I | etcdserver/api: enabled capabilities for version 3.2
```
```
$ ETCDCTL_API=3 /etcdctl endpoint health --endpoints=localhost:2379,localhost:22379,localhost:32379
localhost:2379 is healthy: successfully committed proposal: took = 2.312897ms
localhost:22379 is healthy: successfully committed proposal: took = 2.553476ms
localhost:32379 is healthy: successfully committed proposal: took = 2.517902ms
In the general case, upgrading from etcd 3.2 to 3.3 can be a zero-downtime, rolling upgrade:
- one by one, stop the etcd v3.2 processes and replace them with etcd v3.3 processes
- after running all v3.3 processes, new features in v3.3 are available to the cluster
Before [starting an upgrade](#upgrade-procedure), read through the rest of this guide to prepare.
### Client upgrade checklists
3.3 introduces breaking changes (TODO: update this before 3.3 release).
Previously, `grpc.ErrClientConnTimeout` error is returned on client dial time-outs. 3.3 instead returns `context.DeadlineExceeded` (see [#8504](https://github.com/coreos/etcd/issues/8504)).
Before
```go
// expect dial time-out on ipv4 blackhole
_,err:=clientv3.New(clientv3.Config{
Endpoints:[]string{"http://254.0.0.1:12345"},
DialTimeout:2*time.Second
})
iferr==grpc.ErrClientConnTimeout{
// handle errors
}
```
After
```go
_,err:=clientv3.New(clientv3.Config{
Endpoints:[]string{"http://254.0.0.1:12345"},
DialTimeout:2*time.Second
})
iferr==context.DeadlineExceeded{
// handle errors
}
```
### Server upgrade checklists
#### Upgrade requirements
To upgrade an existing etcd deployment to 3.3, the running cluster must be 3.2 or greater. If it's before 3.2, please [upgrade to 3.2](upgrade_3_2.md) before upgrading to 3.3.
Also, to ensure a smooth rolling upgrade, the running cluster must be healthy. Check the health of the cluster by using the `etcdctl endpoint health` command before proceeding.
#### Preparation
Before upgrading etcd, always test the services relying on etcd in a staging environment before deploying the upgrade to the production environment.
Before beginning, [backup the etcd data](../op-guide/maintenance.md#snapshot-backup). Should something go wrong with the upgrade, it is possible to use this backup to [downgrade](#downgrade) back to existing etcd version. Please note that the `snapshot` command only backs up the v3 data. For v2 data, see [backing up v2 datastore](../v2/admin_guide.md#backing-up-the-datastore).
#### Mixed versions
While upgrading, an etcd cluster supports mixed versions of etcd members, and operates with the protocol of the lowest common version. The cluster is only considered upgraded once all of its members are upgraded to version 3.3. Internally, etcd members negotiate with each other to determine the overall cluster version, which controls the reported version and the supported features.
#### Limitations
Note: If the cluster only has v3 data and no v2 data, it is not subject to this limitation.
If the cluster is serving a v2 data set larger than 50MB, each newly upgraded member may take up to two minutes to catch up with the existing cluster. Check the size of a recent snapshot to estimate the total data size. In other words, it is safest to wait for 2 minutes between upgrading each member.
For a much larger total data size, 100MB or more , this one-time process might take even more time. Administrators of very large etcd clusters of this magnitude can feel free to contact the [etcd team][etcd-contact] before upgrading, and we'll be happy to provide advice on the procedure.
#### Downgrade
If all members have been upgraded to v3.3, the cluster will be upgraded to v3.3, and downgrade from this completed state is **not possible**. If any single member is still v3.2, however, the cluster and its operations remains "v3.2", and it is possible from this mixed cluster state to return to using a v3.2 etcd binary on all members.
Please [backup the data directory](../op-guide/maintenance.md#snapshot-backup) of all etcd members to make downgrading the cluster possible even after it has been completely upgraded.
### Upgrade procedure
This example shows how to upgrade a 3-member v3.2 ectd cluster running on a local machine.
#### 1. Check upgrade requirements
Is the cluster healthy and running v3.2.x?
```
$ ETCDCTL_API=3 etcdctl endpoint health --endpoints=localhost:2379,localhost:22379,localhost:32379
localhost:2379 is healthy: successfully committed proposal: took = 6.600684ms
localhost:22379 is healthy: successfully committed proposal: took = 8.540064ms
localhost:32379 is healthy: successfully committed proposal: took = 8.763432ms
$ curl http://localhost:2379/version
{"etcdserver":"3.2.7","etcdcluster":"3.2.0"}
```
#### 2. Stop the existing etcd process
When each etcd process is stopped, expected errors will be logged by other cluster members. This is normal since a cluster member connection has been (temporarily) broken:
```
14:13:31.491746 I | raft: c89feb932daef420 [term 3] received MsgTimeoutNow from 6d4f535bae3ab960 and starts an election to get leadership.
14:13:31.491769 I | raft: c89feb932daef420 became candidate at term 4
14:13:31.491788 I | raft: c89feb932daef420 received MsgVoteResp from c89feb932daef420 at term 4
14:13:31.491797 I | raft: c89feb932daef420 [logterm: 3, index: 9] sent MsgVote request to 6d4f535bae3ab960 at term 4
14:13:31.491805 I | raft: c89feb932daef420 [logterm: 3, index: 9] sent MsgVote request to 9eda174c7df8a033 at term 4
14:13:31.491815 I | raft: raft.node: c89feb932daef420 lost leader 6d4f535bae3ab960 at term 4
14:13:31.524084 I | raft: c89feb932daef420 received MsgVoteResp from 6d4f535bae3ab960 at term 4
14:13:31.524108 I | raft: c89feb932daef420 [quorum:2] has received 2 MsgVoteResp votes and 0 vote rejections
14:13:31.524123 I | raft: c89feb932daef420 became leader at term 4
14:13:31.524136 I | raft: raft.node: c89feb932daef420 elected leader c89feb932daef420 at term 4
14:13:31.592650 W | rafthttp: lost the TCP streaming connection with peer 6d4f535bae3ab960 (stream MsgApp v2 reader)
14:13:31.592825 W | rafthttp: lost the TCP streaming connection with peer 6d4f535bae3ab960 (stream Message reader)
14:13:31.693275 E | rafthttp: failed to dial 6d4f535bae3ab960 on stream Message (dial tcp [::1]:2380: getsockopt: connection refused)
14:13:31.693289 I | rafthttp: peer 6d4f535bae3ab960 became inactive
14:13:31.936678 W | rafthttp: lost the TCP streaming connection with peer 6d4f535bae3ab960 (stream Message writer)
```
It's a good idea at this point to [backup the etcd data](../op-guide/maintenance.md#snapshot-backup) to provide a downgrade path should any problems occur:
```
$ etcdctl snapshot save backup.db
```
#### 3. Drop-in etcd v3.3 binary and start the new etcd process
The new v3.3 etcd will publish its information to the cluster:
```
14:14:25.363225 I | etcdserver: published {Name:s1 ClientURLs:[http://localhost:2379]} to cluster a9ededbffcb1b1f1
```
Verify that each member, and then the entire cluster, becomes healthy with the new v3.3 etcd binary:
```
$ ETCDCTL_API=3 /etcdctl endpoint health --endpoints=localhost:2379,localhost:22379,localhost:32379
localhost:22379 is healthy: successfully committed proposal: took = 5.540129ms
localhost:32379 is healthy: successfully committed proposal: took = 7.321771ms
localhost:2379 is healthy: successfully committed proposal: took = 10.629901ms
```
Upgraded members will log warnings like the following until the entire cluster is upgraded. This is expected and will cease after all etcd cluster members are upgraded to v3.3:
```
14:15:17.071804 W | etcdserver: member c89feb932daef420 has a higher version 3.3.0
14:15:21.073110 W | etcdserver: the local etcd version 3.2.7 is not up-to-date
14:15:21.073142 W | etcdserver: member 6d4f535bae3ab960 has a higher version 3.3.0
14:15:21.073157 W | etcdserver: the local etcd version 3.2.7 is not up-to-date
14:15:21.073164 W | etcdserver: member c89feb932daef420 has a higher version 3.3.0
```
#### 4. Repeat step 2 to step 3 for all other members
#### 5. Finish
When all members are upgraded, the cluster will report upgrading to 3.3 successfully:
```
14:15:54.536901 N | etcdserver/membership: updated the cluster version from 3.2 to 3.3
14:15:54.537035 I | etcdserver/api: enabled capabilities for version 3.3
```
```
$ ETCDCTL_API=3 /etcdctl endpoint health --endpoints=localhost:2379,localhost:22379,localhost:32379
localhost:2379 is healthy: successfully committed proposal: took = 2.312897ms
localhost:22379 is healthy: successfully committed proposal: took = 2.553476ms
localhost:32379 is healthy: successfully committed proposal: took = 2.517902ms
**This is the documentation for etcd2 releases. Read [etcd3 doc][v3-docs] for etcd3 releases.**
[v3-docs]: ../docs.md#documentation
# Snapshot Migration
You can migrate a snapshot of your data from a v0.4.9+ cluster into a new etcd 2.2 cluster using a snapshot migration. After snapshot migration, the etcd indexes of your data will change. Many etcd applications rely on these indexes to behave correctly. This operation should only be done while all etcd applications are stopped.
[](https://quay.io/repository/coreos/etcd-git)
etcd is a distributed key-value store designed to reliably and quickly preserve and provide access to critical data. It enables reliable distributed coordination through distributed locking, leader elections, and write barriers. An etcd cluster is intended for high availability and permanent data storage and retrieval.
**Note**: The `master` branch may be in an *unstable or even broken state* during development. Please use [releases][github-release] instead of the `master` branch in order to get stable binaries.
This is the etcd v2 documentation set. For more recent versions, please see the [etcd v3 guides][etcd-v3].
etcd is a distributed, consistent key-value store for shared configuration and service discovery, with a focus on being:
Reading and writing into the etcd keyspace is done via a simple, RESTful HTTP API, or using language-specific libraries that wrap the HTTP API with higher level primitives.
* *Fast*: benchmarked 1000s of writes/s per instance
* *Reliable*: properly distributed using Raft
### Reading and Writing
etcd is written in Go and uses the [Raft][raft] consensus algorithm to manage a highly-available replicated log.
- [Client API Documentation][api]
- [Libraries, Tools, and Language Bindings][libraries]
- [Admin API Documentation][admin-api]
- [Members API][members-api]
etcd is used [in production by many companies](./production-users.md), and the development team stands behind it in critical deployment scenarios, where etcd is frequently teamed with applications such as [Kubernetes][k8s], [fleet][fleet], [locksmith][locksmith], [vulcand][vulcand], and many others.
### Security, Auth, Access control
See [etcdctl][etcdctl] for a simple command line client.
Or feel free to just use `curl`, as in the examples below.
Configuration values are distributed within the cluster for your applications to read. Values can be changed programmatically and smart applications can reconfigure automatically. You'll never again have to run a configuration management tool on every machine in order to change a single config value.
### Getting etcd
### General Info
The easiest way to get etcd is to use one of the pre-built release binaries which are available for OSX, Linux, Windows, AppC (ACI), and Docker. Instructions for using these binaries are on the [GitHub releases page][github-release].
- [etcd Proxies][proxy]
- [Production Users][production-users]
- [Admin Guide][admin_guide]
- [Configuration Flags][configuration]
- [Frequently Asked Questions][faq]
For those wanting to try the very latest version, you can build the latest version of etcd from the `master` branch.
You will first need [*Go*](https://golang.org/) installed on your machine (version 1.5+ is required).
All development occurs on `master`, including new features and bug fixes.
Bug fixes are first targeted at `master` and subsequently ported to release branches, as described in the [branch management][branch-management] guide.
You have successfully started an etcd and written a key to the store.
### etcd TCP ports
The [official etcd ports][iana-ports] are 2379 for client requests, and 2380 for peer communication. To maintain compatibility, some etcd configuration and documentation continues to refer to the legacy ports 4001 and 7001, but all new etcd use and discussion should adopt the IANA-assigned ports. The legacy ports 4001 and 7001 will be fully deprecated, and support for their use removed, in future etcd releases.
If the previous node is a key and client tries to overwrite it with `dir=true`, it does not give warnings such as `Not a directory`. Instead, the key is set to empty value.
**This is the documentation for etcd2 releases. Read [etcd3 doc][v3-docs] for etcd3 releases.**
[v3-docs]: ../docs.md#documentation
# Administration
## Data Directory
@ -8,7 +13,7 @@ When first started, etcd stores its configuration into a data directory specifie
Configuration is stored in the write ahead log and includes: the local member ID, cluster ID, and initial cluster configuration.
The write ahead log and snapshot files are used during member operation and to recover after a restart.
Having a dedicated disk to store wal files can improve the throughput and stabilize the cluster.
Having a dedicated disk to store wal files can improve the throughput and stabilize the cluster.
It is highly recommended to dedicate a wal disk and set `--wal-dir` to point to a directory on that device for a production cluster deployment.
If a member’s data directory is ever lost or corrupted then the user should [remove][remove-a-member] the etcd member from the cluster using `etcdctl` tool.
@ -40,18 +45,18 @@ It is important to monitor your production etcd cluster for healthy information
#### Health Monitoring
At lowest level, etcd exposes health information via HTTP at `/health` in JSON format. If it returns `{"health": "true"}`, then the cluster is healthy. Please note the `/health` endpoint is still an experimental one as in etcd 2.2.
At lowest level, etcd exposes health information via HTTP at `/health` in JSON format. If it returns `{"health":true}`, then the cluster is healthy.
```
$ curl -L http://127.0.0.1:2379/health
{"health": "true"}
{"health":true}
```
You can also use etcdctl to check the cluster-wide health information. It will contact all the members of the cluster and collect the health information for you.
```
$./etcdctl cluster-health
$./etcdctl cluster-health
member 8211f1d0f64f3269 is healthy: got healthy result from http://127.0.0.1:12379
member 91bc3c398fb3c146 is healthy: got healthy result from http://127.0.0.1:22379
member fd422379fda50e48 is healthy: got healthy result from http://127.0.0.1:32379
**This is the documentation for etcd2 releases. Read [etcd3 doc][v3-docs] for etcd3 releases.**
[v3-docs]: ../docs.md#documentation
# etcd3 API
TODO: API doc
@ -47,7 +52,7 @@ An etcd operation is considered complete when it is committed through consensus,
#### revision
An etcd operation that modifies the key value store is assigned with a single increasing revision. A transaction operation might modifies the key value store multiple times, but only one revision is assigned. The revision attribute of a key value pair that modified by the operation has the same value as the revision of the operation. The revision can be used as a logical clock for key value store. A key value pair that has a larger revision is modified after a key value pair with a smaller revision. Two key value pairs that have the same revision are modified by an operation "concurrently".
An etcd operation that modifies the key value store is assigned with a single increasing revision. A transaction operation might modify the key value store multiple times, but only one revision is assigned. The revision attribute of a key value pair that modified by the operation has the same value as the revision of the operation. The revision can be used as a logical clock for key value store. A key value pair that has a larger revision is modified after a key value pair with a smaller revision. Two key value pairs that have the same revision are modified by an operation "concurrently".
A user is an identity to be authenticated. Each user can have multiple roles. The user has a capability (such as reading or writing) on the resource if one of the roles has that capability.
@ -15,7 +20,7 @@ A user is an identity to be authenticated. Each user can have multiple roles. Th
A user named `root` is required before authentication can be enabled, and it always has the ROOT role. The ROOT role can be granted to multiple users, but `root` is required for recovery purposes.
#### Roles
Each role has exact one associated Permission List. An permission list exists for each permission on key-value resources.
Each role has exact one associated Permission List. An permission list exists for each permission on key-value resources.
The special static ROOT (named `root`) role has a full permissions on all key-value resources, the permission to manage user resources and settings resources. Only the ROOT role has the permission to manage user resources and modify settings resources. The ROOT role is built-in and does not need to be created.
@ -30,8 +35,8 @@ A Permission List is a list of allowed patterns for that particular permission (
### Key-Value Resources
A key-value resource is a key-value pairs in the store. Given a list of matching patterns, permission for any given key in a request is granted if any of the patterns in the list match.
Only prefixes or exact keys are supported. A prefix permission string ends in `*`.
A permission on `/foo` is for that exact key or directory, not its children or recursively. `/foo*` is a prefix that matches `/foo` recursively, and all keys thereunder, and keys with that prefix (eg. `/foobar`. Contrast to the prefix `/foo/*`). `*` alone is permission on the full keyspace.
Only prefixes or exact keys are supported. A prefix permission string ends in `*`.
A permission on `/foo` is for that exact key or directory, not its children or recursively. `/foo*` is a prefix that matches `/foo` recursively, and all keys thereunder, and keys with that prefix (eg. `/foobar`. Contrast to the prefix `/foo/*`). `*` alone is permission on the full keyspace.
### Settings Resources
@ -66,7 +71,7 @@ An Error JSON corresponds to:
}
#### Enable and Disable Authentication
**Get auth status**
GET /v2/auth/enable
@ -215,8 +220,8 @@ PUT /v2/auth/users/charlie
Sent Headers:
Authorization: Basic <BasicAuthString>
Put Body:
JSON struct, above, matching the appropriate name
* Starting password and roles when creating.
JSON struct, above, matching the appropriate name
* Starting password and roles when creating.
* Grant/Revoke/Password filled in when updating (to grant roles, revoke roles, or change the password).
Possible Status Codes:
200 OK
@ -345,7 +350,7 @@ PUT /v2/auth/roles/rkt
401 Unauthorized
404 Not Found (update non-existent roles)
409 Conflict (when granting duplicated permission or revoking non-existent permission)
**This is the documentation for etcd2 releases. Read [etcd3 doc][v3-docs] for etcd3 releases.**
[v3-docs]: ../docs.md#documentation
# Authentication Guide
## Overview
@ -14,7 +19,7 @@ There is one special user, `root`, and there are two special roles, `root` and `
### User `root`
User `root` must be created before security can be activated. It has the `root` role and allows for the changing of anything inside etcd. The idea behind the `root` user is for recovery purposes -- a password is generated and stored somewhere -- and the root role is granted to the administrator accounts on the system. In the future, for troubleshooting and recovery, we will need to assume some access to the system, and future documentation will assume this root user (though anyone with the role will suffice).
User `root` must be created before security can be activated. It has the `root` role and allows for the changing of anything inside etcd. The idea behind the `root` user is for recovery purposes -- a password is generated and stored somewhere -- and the root role is granted to the administrator accounts on the system. In the future, for troubleshooting and recovery, we will need to assume some access to the system, and future documentation will assume this root user (though anyone with the role will suffice).
### Role `root`
@ -104,7 +109,7 @@ $ etcdctl role grant myrolename -path '/foo/bar' -write
$ etcdctl role grant myrolename -path '/pub/*' -readwrite
```
Beware that
Beware that
```
# Give full access to keys under /pub??
@ -133,12 +138,12 @@ $ etcdctl role remove myrolename
## Enabling authentication
The minimal steps to enabling auth are as follows. The administrator can set up users and roles before or after enabling authentication, as a matter of preference.
The minimal steps to enabling auth are as follows. The administrator can set up users and roles before or after enabling authentication, as a matter of preference.
**This is the documentation for etcd2 releases. Read [etcd3 doc][v3-docs] for etcd3 releases.**
[v3-docs]: ../docs.md#documentation
# Backward Compatibility
The main goal of etcd 2.0 release is to improve cluster safety around bootstrapping and dynamic reconfiguration. To do this, we deprecated the old error-prone APIs and provide a new set of APIs.
**This is the documentation for etcd2 releases. Read [etcd3 doc][v3-docs] for etcd3 releases.**
[v3-docs]: ../../docs.md#documentation
# Watch Memory Usage Benchmark
*NOTE*: The watch features are under active development, and their memory usage may change as that development progresses. We do not expect it to significantly increase beyond the figures stated below.
@ -5,10 +10,10 @@
A primary goal of etcd is supporting a very large number of watchers doing a massively large amount of watching. etcd aims to support O(10k) clients, O(100K) watch streams (O(10) streams per client) and O(10M) total watchings (O(100) watching per stream). The memory consumed by each individual watching accounts for the largest portion of etcd's overall usage, and is therefore the focus of current and future optimizations.
Three related components of etcd watch consume physical memory: each `grpc.Conn`, each watch stream, and each instance of the watching activity. `grpc.Conn` maintains the actual TCP connection and other gRPC connection state. Each `grpc.Conn` consumes O(10kb) of memory, and might have multiple watch streams attached.
Three related components of etcd watch consume physical memory: each `grpc.Conn`, each watch stream, and each instance of the watching activity. `grpc.Conn` maintains the actual TCP connection and other gRPC connection state. Each `grpc.Conn` consumes O(10kb) of memory, and might have multiple watch streams attached.
Each watch stream is an independent HTTP2 connection which consumes another O(10kb) of memory.
Multiple watchings might share one watch stream.
Each watch stream is an independent HTTP2 connection which consumes another O(10kb) of memory.
Multiple watchings might share one watch stream.
Watching is the actual struct that tracks the changes on the key-value store. Each watching should only consume <O(1kb).
**This is the documentation for etcd2 releases. Read [etcd3 doc][v3-docs] for etcd3 releases.**
[v3-docs]: ../../docs.md#documentation
# Storage Memory Usage Benchmark
<!---todo: link storage to storage design doc-->
@ -60,7 +65,7 @@ GCE n1-standard-2 machine type
In this test, we only benchmark the memory usage of the in-memory index. The goal is to find `c1` and `c2` mentioned above and to understand the hard limit of memory consumption of the storage.
We calculate the memory usage consumption via the Go runtime.ReadMemStats. We calculate the total allocated bytes difference before creating the index and after creating the index. It cannot perfectly reflect the memory usage of the in-memory index itself but can show the rough consumption pattern.
We calculate the memory usage consumption via the Go runtime.ReadMemStats. We calculate the total allocated bytes difference before creating the index and after creating the index. It cannot perfectly reflect the memory usage of the in-memory index itself but can show the rough consumption pattern.
**This is the documentation for etcd2 releases. Read [etcd3 doc][v3-docs] for etcd3 releases.**
[v3-docs]: ../docs.md#documentation
# Configuration Flags
etcd is configurable through command-line flags and environment variables. Options set on the command line take precedence over those from the environment.
@ -176,7 +181,10 @@ To start etcd automatically using custom settings at startup in Linux, using a [
The security flags help to [build a secure etcd cluster][security].
### --ca-file [DEPRECATED]
### --ca-file
**DEPRECATED**
+ Path to the client server TLS CA file. `--ca-file ca.crt` could be replaced by `--trusted-ca-file ca.crt --client-cert-auth` and etcd will perform the same.
+ default: none
+ env variable: ETCD_CA_FILE
@ -197,11 +205,14 @@ The security flags help to [build a secure etcd cluster][security].
+ env variable: ETCD_CLIENT_CERT_AUTH
### --trusted-ca-file
+ Path to the client server TLS trusted CA key file.
+ Path to the client server TLS trusted CA cert file.
+ default: none
+ env variable: ETCD_TRUSTED_CA_FILE
### --peer-ca-file [DEPRECATED]
### --peer-ca-file
**DEPRECATED**
+ Path to the peer server TLS CA file. `--peer-ca-file ca.crt` could be replaced by `--peer-trusted-ca-file ca.crt --peer-client-cert-auth` and etcd will perform the same.
**This is the documentation for etcd2 releases. Read [etcd3 doc][v3-docs] for etcd3 releases.**
[v3-docs]: ../../docs.md#documentation
# etcd release guide
The guide talks about how to release a new version of etcd.
The procedure includes some manual steps for sanity checking but it can probably be further scripted. Please keep this document up-to-date if you want to make changes to the release process.
The procedure includes some manual steps for sanity checking but it can probably be further scripted. Please keep this document up-to-date if you want to make changes to the release process.
## Prepare Release
@ -70,7 +75,7 @@ cd release
# personal GPG is okay for now
for i in etcd-*{.zip,.tar.gz}; do gpg --sign ${i}; done
# use `CoreOS ACI Builder <release@coreos.com>` secret key
gpg -u 88182190 -a --output etcd-${VERSION}-linux-amd64.aci.asc --detach-sig etcd-${VERSION}-linux-amd64.aci
for aci in etcd-${VERSION}.*.aci; do gpg -u 88182190 -a --output ${aci}.asc --detach-sig ${aci}; done
@ -42,11 +47,13 @@ etcdctl -C http://192.168.12.50:4001 member list
Using Docker to setup a multi-node cluster is very similar to the standalone mode configuration.
The main difference being the value used for the `-initial-cluster` flag, which must contain the peer urls for each etcd member in the cluster.
**Although the following commands look very similar, note that `-name`, `-advertise-client-urls` and `-initial-advertise-peer-urls` differ for each cluster member**
**This is the documentation for etcd2 releases. Read [etcd3 doc][v3-docs] for etcd3 releases.**
[v3-docs]: ../docs.md#documentation
# Metrics
etcd uses [Prometheus][prometheus] for metrics reporting. The metrics can be used for real-time monitoring and debugging. etcd does not persist its metrics; if a member restarts, the metrics will be reset.
@ -14,9 +19,9 @@ The metrics under the `etcd` prefix are for monitoring and alerting. They are st
### http requests
These metrics describe the serving of requests (non-watch events) served by etcd members in non-proxy mode: total
These metrics describe the serving of requests (non-watch events) served by etcd members in non-proxy mode: total
incoming requests, request failures and processing latency (inc. raft rounds for storage). They are useful for tracking
user-generated traffic hitting the etcd cluster .
user-generated traffic hitting the etcd cluster .
All these metrics are prefixed with `etcd_http_`
@ -28,20 +33,20 @@ All these metrics are prefixed with `etcd_http_`
Example Prometheus queries that may be useful from these metrics (across all etcd members):
*`sum(rate(etcd_http_failed_total{job="etcd"}[1m]) by (method) / sum(rate(etcd_http_events_received_total{job="etcd"})[1m]) by (method)`
*`sum(rate(etcd_http_failed_total{job="etcd"}[1m]) by (method) / sum(rate(etcd_http_events_received_total{job="etcd"})[1m]) by (method)`
Shows the fraction of events that failed by HTTP method across all members, across a time window of `1m`.
*`sum(rate(etcd_http_received_total{job="etcd",method="GET})[1m]) by (method)`
`sum(rate(etcd_http_received_total{job="etcd",method~="GET})[1m]) by (method)`
Shows the rate of successful readonly/write queries across all servers, across a time window of `1m`.
*`histogram_quantile(0.9, sum(rate(etcd_http_successful_duration_seconds{job="etcd",method="GET"}[5m]) ) by (le))`
`histogram_quantile(0.9, sum(rate(etcd_http_successful_duration_seconds{job="etcd",method!="GET"}[5m]) ) by (le))`
Show the 0.90-tile latency (in seconds) of read/write (respectively) event handling across all members, with a window of `5m`.
Show the 0.90-tile latency (in seconds) of read/write (respectively) event handling across all members, with a window of `5m`.
### proxy
@ -56,21 +61,21 @@ All these metrics are prefixed with `etcd_proxy_`
| requests_total | Total number of requests by this proxy instance. | Counter(method) |
| handled_total | Total number of fully handled requests, with responses from etcd members. | Counter(method) |
| dropped_total | Total number of dropped requests due to forwarding errors to etcd members. | Counter(method,error) |
| handling_duration_seconds | Bucketed handling times by HTTP method, including round trip to member instances. | Histogram(method) |
| handling_duration_seconds | Bucketed handling times by HTTP method, including round trip to member instances. | Histogram(method) |
Example Prometheus queries that may be useful from these metrics (across all etcd servers):
*`sum(rate(etcd_proxy_handled_total{job="etcd"}[1m])) by (method)`
Rate of requests (by HTTP method) handled by all proxies, across a window of `1m`.
Rate of requests (by HTTP method) handled by all proxies, across a window of `1m`.
*`histogram_quantile(0.9, sum(rate(handling_duration_seconds{job="etcd",method="GET"}[5m])) by (le))`
`histogram_quantile(0.9, sum(rate(handling_duration_seconds{job="etcd",method!="GET"}[5m])) by (le))`
Show the 0.90-tile latency (in seconds) of handling of user requests across all proxy machines, with a window of `5m`.
Show the 0.90-tile latency (in seconds) of handling of user requests across all proxy machines, with a window of `5m`.
*`sum(rate(etcd_proxy_dropped_total{job="etcd"}[1m])) by (proxying_error)`
Number of failed request on the proxy. This should be 0, spikes here indicate connectivity issues to the etcd cluster.
**This is the documentation for etcd2 releases. Read [etcd3 doc][v3-docs] for etcd3 releases.**
[v3-docs]: ../docs.md#documentation
# Production Users
This document tracks people and use cases for etcd in production. By creating a list of production use cases we hope to build a community of advisors that we can reach out to with experience using various etcd applications, operation environments, and cluster sizes. The etcd development team may reach out periodically to check-in on your experience and update this list.
**This is the documentation for etcd2 releases. Read [etcd3 doc][v3-docs] for etcd3 releases.**
[v3-docs]: ../docs.md#documentation
# Proxy
etcd can run as a transparent proxy. Doing so allows for easy discovery of etcd within your infrastructure, since it can run on each machine as a local service. In this mode, etcd acts as a reverse proxy and forwards client requests to an active etcd cluster. The etcd proxy does not participate in the consensus replication of the etcd cluster, thus it neither increases the resilience nor decreases the write performance of the etcd cluster.
**This is the documentation for etcd2 releases. Read [etcd3 doc][v3-docs] for etcd3 releases.**
[v3-docs]: ../docs.md#documentation
# Reporting Bugs
If you find bugs or documentation mistakes in the etcd project, please let us know by [opening an issue][etcd-issue]. We treat bugs and mistakes very seriously and believe no issue is too small. Before creating a bug report, please check that an issue reporting the same problem does not already exist.
Some files were not shown because too many files have changed in this diff
Show More
Reference in New Issue
Block a user
Blocking a user prevents them from interacting with repositories, such as opening or commenting on pull requests or issues. Learn more about blocking a user.
