This attempts to fix a special case of the problem described in #12385,
where trying to do `clientv3.Watch` with an expired token would result
in `ErrGRPCPermissionDenied`, due to the failing authorization check in
`isWatchPermitted`. Furthermore, the client can't auto recover, since
`shouldRefreshToken` rightly returns false for the permission denied
error.
In this case, we would like to have a runbook to dynamically disable
auth, without causing any disruption. Doing so would immediately expire
all existing tokens, which would then cause the behavior described
above. This means existing watchers would still work for a period of
time after disabling auth, until they have to reconnect, e.g. due to a
rolling restart of server nodes.
This commit adds a client-side fix and a server-side fix, either of
which is sufficient to get the added test case to pass. Note that it is
an e2e test case instead of an integration one, as the reconnect only
happens if the server node is stopped via SIGINT or SIGTERM.
A generic fix for the problem described in #12385 would be better, as
that shall also fix this special case. However, the fix would likely be
a lot more involved, as some untangling of authn/authz is required.
In the sample code demonstrating how to specify a client request
timeout, the `cancel()` is called immediately after the Put, but it
should be deferred instead, giving the Put enough time to complete.
In the canonical Golang context
[docs](https://pkg.go.dev/context#WithTimeout), the sample code sets a
`defer cancel()` immediately after context creation, and with this
commit we adhere to that convention.
fixes:
```json
{
"level": "warn",
"ts": "2021-12-29T09:56:42.439-0800",
"logger": "etcd-client",
"caller": "v3@v3.5.1/retry_interceptor.go:62",
"msg": "retrying of unary invoker failed",
"target": "etcd-endpoints://0xc000213340/localhost:2379",
"attempt": 0,
"error": "rpc error: code = Canceled desc = context canceled"
}
```
raft: fix goroutine leaks in TestCommitPagination
The goroutine created with n.run() will leak if we forget to call n.Stop().
We can replay the goroutine leaks by using [goleak](https://github.com/uber-go/goleak):
```
$ cd raft && env go test -short -v -timeout=3m --race -run=TestCommitPagination.
... ...
raft2021/12/27 20:47:15 INFO: raft.node: 1 elected leader 1 at term 1
leaks.go:78: found unexpected goroutines:
[Goroutine 20 in state select, with go.etcd.io/etcd/raft/v3.(*node).run on top of the stack:
goroutine 20 [select]:
go.etcd.io/etcd/raft/v3.(*node).run(0xc00036f260)
/home/yuanting/work/dev/goprojects/etcd/raft/node.go:344 +0xc1d
created by go.etcd.io/etcd/raft/v3.TestCommitPagination
/home/yuanting/work/dev/goprojects/etcd/raft/node_test.go:920 +0x554
]
--- FAIL: TestCommitPagination (0.45s)
FAIL
FAIL go.etcd.io/etcd/raft/v3 0.508s
FAIL
```
This has been additionally verified by running the tests locally as a
basic smoke test. GitHub Actions doesn't provide MacOS M1 (arm64) yet,
so there's no good way to automate testing.
Ran `TMPDIR=/tmp make test` locally. The `TMPDIR` bit is needed so
there's no really long path used that breaks Unix socket setup in one of
the tests.
