[3.4] Test failure: "Too many goroutines running after all test"

[ahrtr]

We are going to release 3.4.19, and the top priority is to make the pipeline as stable as possible.

Too many goroutines running after all test(s).
1 instances of:
go.etcd.io/etcd/etcdserver/api/v3rpc.(*serverWatchStream).recvLoop(...)
	/home/runner/work/etcd/etcd/etcdserver/api/v3rpc/watch.go:336 +0x214
go.etcd.io/etcd/etcdserver/api/v3rpc.(*watchServer).Watch.func2(...)
	/home/runner/work/etcd/etcd/etcdserver/api/v3rpc/watch.go:190 +0x45
created by go.etcd.io/etcd/etcdserver/api/v3rpc.(*watchServer).Watch
	/home/runner/work/etcd/etcd/etcdserver/api/v3rpc/watch.go:189 +0x2ec
FAIL	go.etcd.io/etcd/clientv3/integration	270.490s
FAIL

Refer to https://github.com/etcd-io/etcd/runs/7049120619?check_suite_focus=true

[lavacat]

/assign

[ahrtr]

@lavacat Do you have any update on this ticket?

[lavacat]

no updates yet, only looked at the other one

[lavacat]

I wasn't able to reproduce this and I went through GitHub Actions runs for ~10days and didn't see this error. The most relevant comment about similar issue here

I think there might be 'false negatives', i.e. situation when goroutine is 'RUNNING' (not waiting), so its not reported by list of goroutines, so we might exit leak detection when there is actually a goroutine leaked. But it should not be reported as failure.

I suggest we close this out

[ahrtr]

Thanks, let's reopen it once we see it again.

[MadhavJivrajani]

https://github.com/etcd-io/etcd/actions/runs/9093821374/job/25036682192?pr=18016

Saw an instance of this again ^