Closed yutkin closed 11 months ago
Thank you for reporting this issue. It looks like https://github.com/grpc/grpc-go/pull/6489 might be the culprit, but I'm not certain at this point. Needs more investigation.
Do you have any way to reproduce this?
Also, do you use the NumStreamWorkers
server option?
Do you have any way to reproduce this?
We don't now, unfortunately. We've just reverted this when we noticed the issue.
Also, do you use the NumStreamWorkers server option?
Yes, we do. We set this the same as GOMAXPROCS and pod's CPU requests/limits.
Are the panics happening around the time when Stop
/GracefulStop
is being called on your grpc server?
Are the panics happening around the time when
Stop
/GracefulStop
is being called on your grpc server?
Yes, they are happening right after the GracefulStop
What version of gRPC are you using?
1.60
What version of Go are you using (
go version
)?1.21.3
What operating system (Linux, Windows, …) and version?
cos_containerd on GKE
What did you do?
After upgrading from 1.59 to 1.60 we started to get panics from time to time. We didn't change anything else.
We use Proxyless Traffic Director in GCP.
What did you expect to see?
What did you see instead?