notnoopci / bazel-remote-proxy

MIT License
38 stars 5 forks source link

remote cache stopped working a few days ago #2

Open Globegitter opened 5 years ago

Globegitter commented 5 years ago

A few days ago we started seeing error messages from bazel-remote-proxy and the remote cache stopped working:

WARN[0157] error retrieving storage credentials          error="rpc error: code = Unimplemented desc = unknown service protocol.EventProcessor"
WARN[0157] error fetching storage credentials, retrying  error="rpc error: code = Unimplemented desc = unknown service protocol.EventProcessor"

Is there anything that changed on the circleci side? Anything that needs to be updated here?

Globegitter commented 5 years ago

@notnoopci gentle ping - any idea about this?

marcesquerra commented 5 years ago

We are having the same issue. Is there anything we can do to try and help solve this?

Globegitter commented 5 years ago

It seems the error is coming from: https://github.com/notnoopci/bazel-remote-proxy/blob/a07b0b3b7935d95a24e947832503d09c7397c35a/circleci/credsprovider.go#L32 as the backoff is happening. IT seems to me like the grpc service has changed and things need to be updated here. So my guess is that https://github.com/notnoopci/bazel-remote-proxy/blob/master/circleci/protocol/pb.go is the file that needs to be updated as well as any proto calls. But not sure where the source proto file is located.

meastham commented 5 years ago

I'm having this issue as well. Did anybody figure out where that proto definition is coming from?

Globegitter commented 5 years ago

My gut feeling is they probably got copied out from an internal repository and would need to be updated by a circleci employee. Maybe worth trying to get in touch with circleci?

kamalmarhubi commented 5 years ago

I tried to use gRPC reflection to query the internal server, but I don't think they're running it with reflection enabled. Was hopeful that'd get a fresh proto definition to rebuild from. Oh well.

kamalmarhubi commented 4 years ago

@notnoopci checking in on this issue again. Any chance of an update to reflect changes to the internal API?