yangwwei opened a new pull request #246: URL: https://github.com/apache/incubator-yunikorn-k8shim/pull/246
Credits to Chaoran who finds this bug and figured out the root cause. The symptom of this issue is described in this issue: https://issues.apache.org/jira/browse/YUNIKORN-584. The problem is when we list applications, we have a filter that checks label `applicationId` exists in pods, pods without this label will be excluded in this scan because [these lines.](https://github.com/apache/incubator-yunikorn-k8shim/blob/fd69f44d0028d09918715cd222775901400f936d/pkg/appmgmt/general/general.go#L297-L302). When we recover Spark jobs, spark pods do not necessarily have this label, we can recognize them using spark-app-selector. This caused the job to get excluded from the recovery app list, and then subsequent node recovery also failed because it finds the existing allocation for the job but the job was not recovered in the scheduler. -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: [email protected]
