Closed kevinpollet closed 4 years ago
This is due to the introduction of https://github.com/containous/maesh/blob/master/integration/try/try.go#L66
Where unavailable replicas are checked to ensure that deployments have rolled over.
If a terminating replica is slow, its possible that this could cause this condition to fail.
Perhaps >0
is not the right condition.
The issue it was implemented to solve was that a deployment was updated, but then instantly checked, and it was still considered "ready", even though it still had to restart all pods, as all current (old) pods were ready.
I am not sure how to improve that condition without checking the pod SHAs
Perhaps we should follow down and check the new ReplicaSet instead of trying to do it at the deployment level.
Bug Report
What did you do?
Opened a PR without modifying the Go code.
What did you expect to see?
Successful build.
What did you see instead?
The integration tests failed because of the
ACLEnabledSuite
as shown here: https://containous.semaphoreci.com/jobs/8bb8e781-dcd1-4292-9f40-a6774ce646ad.A force push without any modifications, fixed the build.