Commit Graph

227 Commits

Author SHA1 Message Date
2cbf7cf6d1 storage: do not send outdated events to unsynced watchers 2016-02-17 14:03:51 -08:00
ee1a03167d storage, v3: pass compaction revision through watchresponse 2016-02-12 14:06:59 -08:00
30c11c1bca *: support local range request 2016-02-12 12:04:06 -08:00
9cd45312d5 Merge pull request #4505 from heyitsanthony/v3-range-ge
storage: support ranges for >= key
2016-02-12 10:03:45 -08:00
4854d7f69d storage: support ranges for >= key
If end == "", range(key, end) will give all keys >= key.
2016-02-12 09:45:43 -08:00
c27a4a1d3d kv.proto: docs of create_revision, mod_revision 2016-02-11 18:25:46 -08:00
40c598cfa6 Merge pull request #4494 from xiang90/sched
*: fix schedule.Wait race
2016-02-11 15:56:01 -08:00
bfa5e310a9 *: detach keys from leases
1. deatch when a key is removed
2. deatch when the key's lease changes
3. potentially deatch when restroing a tombstone key
2016-02-11 15:31:25 -08:00
870e4c2681 *: fix schedule.Wait race 2016-02-11 15:21:33 -08:00
d314345e6d *: add a scheduler and use it to schedule compaction 2016-02-10 14:27:08 -08:00
26c645f049 storage/backend: set initial db size to avoid potential deadlock 2016-02-05 11:29:16 -08:00
b09214df32 storage: update watch.cur and fix tests 2016-02-03 00:54:07 -08:00
52416fafb0 storage: send compaction 2016-02-02 23:17:27 -08:00
5780497e18 storage: remove unncessary handle func 2016-02-02 23:02:15 -08:00
ae5161382b storage: release tx lock until finish using the readonly bytes
The backend will return read only bytes that are only vaild while
the tx is open. We should hold the lock until we get a full copy
by unmarshal.
2016-02-02 22:49:05 -08:00
31c0c5181a storage: make unsync a watcherSetByKey 2016-02-02 20:09:35 -08:00
8dc6248aa7 storage: add set delete 2016-02-02 19:28:42 -08:00
810c3e74a8 storage: remove unnecessary abstraction 2016-02-02 19:15:46 -08:00
e5b35b82c5 storage: add watchSet and watchSetByKey type 2016-02-02 18:56:36 -08:00
20461ab11a *: fix many typos 2016-01-31 21:42:39 -08:00
4ba1ec6a4d Merge pull request #4363 from xiang90/watch
storage: simplify watch store
2016-01-31 16:49:44 -08:00
611751aee2 storage: simplify watch store
We decided that we will not support modifing the same
key in one txn multiple times. That can simlify the current
code/design a lot.
2016-01-31 16:33:07 -08:00
f6215574f2 *: kv range to return current revision
This changes the behavior of KV's range and tx range to return
current revision rather than range revision. This makes populating
range response header easier.
2016-01-30 17:37:34 -08:00
a35d5889f6 *: update gRPC, proto interface 2016-01-26 17:41:39 -08:00
128b5e7387 etcdserver: check invalid range in txn 2016-01-25 20:21:17 -08:00
5a967eb2a0 storage: publish delete events on lease revocation 2016-01-22 13:40:55 -08:00
835d824965 *: minor typos, kill TODOs 2016-01-20 16:21:39 -08:00
ecba9b61cf storage: start initial revision at 1
When the start revision was 0, there was no way to safely watch
starting from the first store revision.
2016-01-14 17:05:56 -08:00
497bbd3010 *: FatalStack to stacktrace tests after timeout
Related to https://github.com/coreos/etcd/issues/4065.
2016-01-14 10:20:14 -08:00
2eac21ae0a Merge pull request #4210 from xiang90/fix_panic
storage: fix panic in test
2016-01-14 08:05:46 -08:00
f2b0689f74 storage: fix panic in test 2016-01-13 23:18:27 -08:00
e1de19bf75 backend: make test more reliable 2016-01-13 22:00:25 -08:00
0cba12d991 storage: extend timeout for slow CI
1. extend timeout

2. print out stacktrace. When it fails again, we can get more confidence that the
failure is caused by slow IO.
2016-01-13 16:04:19 -08:00
4f427bca43 storage: check prefix in unsynced
Current syncWatchers method skips the events that have
prefixes that are being watched when the prefix is not
existent as a key. This fixes https://github.com/coreos/etcd/issues/4191
by adding prefix checking to not skip those events.
2016-01-13 11:21:47 -08:00
f01c8188f8 *: rename lease.DeleteableRange to lease.RangeDeleter 2016-01-09 11:01:58 -08:00
f5753f2f51 *: support lease Attach
Now we can attach keys to leases. And revoking the lease removes all
the attached keys of that lease.
2016-01-09 11:01:58 -08:00
f76166a041 *: fix minor typos 2016-01-08 00:21:19 -08:00
0d01035693 Minor typo in metrics help field 2016-01-07 16:18:50 -05:00
8774d53459 Merge pull request #4158 from heyitsanthony/nolease-to-leasepkg
lease: move storage.NoLease to lease package
2016-01-07 10:43:51 -08:00
f9af744be3 lease: move storage.NoLease to lease package 2016-01-07 10:33:44 -08:00
43a777b7a2 *: get snapshot from backend
We should get snapshot from backend, not just KV.
We plan to store the lease data into backend too.
2016-01-06 22:40:23 -08:00
1714290f4e storage: support recovering from backend
We want the KV to support recovering from backend to avoid
additional pointer swap. Or we have to do coordination between
etcdserver and API layer, since API layer might have access to
kv pointer and use a closed kv.
2016-01-06 21:16:55 -08:00
b546f4c2c2 Merge pull request #4152 from xiang90/fix_force
backend: create bucket should increase pending
2016-01-06 20:47:32 -08:00
5dd8af444a backend: create bucket should increase pending 2016-01-06 20:25:50 -08:00
c70d533771 Merge pull request #4138 from gyuho/watchresponse_header
*: fill in WatchResponse.Header
2016-01-06 15:19:36 -08:00
366e7a879f *: fill in WatchResponse.Header
Related to coreos#3848.
2016-01-06 15:12:53 -08:00
ad29ba3073 storage: remove test file after test 2016-01-05 22:05:28 -08:00
5dd3f91903 *: make backend outside kv
KV and lease will share the same backend. Thus we need to make
backend outside KV.
2016-01-05 19:55:29 -08:00
1e61243fd7 Merge pull request #4134 from xiang90/lease
lease: modify API and persist lease to disk
2016-01-05 10:30:52 -08:00
09b420f08c *: move leaseID typedef to lease pkg 2016-01-05 10:18:17 -08:00