-
Type: Bug
-
Status: Resolved
-
Priority: Minor
-
Resolution: Fixed
-
Affects Version/s: None
-
Component/s: CI/CD
-
Tags:
-
Sprint:nxplatform #37
-
Story Points:3
Randomly, the Kaniko Docker build fails for an unexplained reason:
[2021-06-09T06:07:03.853Z] INFO[5805] cmd: EXPOSE [2021-06-09T06:07:03.853Z] INFO[5805] Adding exposed port: 8080/tcp [2021-06-09T06:07:03.853Z] INFO[5805] ENTRYPOINT ["/docker-entrypoint.sh"] [2021-06-09T06:07:03.853Z] INFO[5805] CMD ["nuxeoctl", "console"] [2021-06-09T06:07:03.853Z] INFO[5805] USER 900 [2021-06-09T06:07:03.853Z] INFO[5805] cmd: USER [2021-06-09T06:09:55.403Z] time="2021-06-09T06:09:51Z" level=fatal msg="build failed: build failed: building [nuxeo]: kaniko build for [nuxeo]: waiting for pod to complete: context closed while waiting for condition" script returned exit code 1
If often happens several times in a row.
It could be due to a CPU/memory resource issue, when the K8s node pool is overloaded.
See the work done in the AI builder to increase the Kaniko resources: https://github.com/nuxeo/jx-ai-builders/blob/master/skaffold.yaml#L26
Or/and we could try these options:
Let's also add tolerations to be tolerated by the Platform dedicated node pool.
Finally, we can try to upgrade to the latest version of Kaniko, it could also help for this issue, and maybe the COPY --chown issue, see NXP-28052.