Closed mmisztal1980 closed 2 years ago
The logs don't seem to show any obvious crash messages from the builder, but the error unable to upgrade connection: container not found (\"buildkitd\")"
seems to imply the container is no longer running.
Let's see if we can gather a little more information from your system to try to understand what's going wrong.
Can you gather the Deployment and all the Pod details in -o yaml
form and paste them into this issue? Hopefully there will be something interesting in the pod events, or other status fields to shed some light on why it stopped working.
Hi,
we've determined that the issues originated from a faulty node with a broken containerd runtime - which caused the pods to get stuck in Terminating
state. (including buildkit).
What steps did you take and what happened We are running a 6-pod buildkit farm. Our CI pipeline contains a step to build & push container images using buildkit:
Today, our developers have started reporting multiple occurences of buildkit cli failing to communicate with the buildkit deployment:
What did you expect to happen We expected our CI step with
kubectl build
to succeedEnvironment Details:
0.1.3
sudo ctr version
or dockerddocker version
on one of your kubernetes worker nodes) - we don't have access to the nodes. we will provide this information when this becomes available. AKS version is 1.19Builder Logs [If applicable, an excerpt from
kubectl logs -l app=buildkit
from around the time you hit the failure may be very helpful]Dockerfile [If applicable, please include your Dockerfile or excerpts related to the failure]
Vote on this request
This is an invitation to the community to vote on issues. Use the "smiley face" up to the right of this comment to vote.