Hi Everyone, We are using K8S operator to manage f...
# troubleshooting
a
Hi Everyone, We are using K8S operator to manage flink Application clusters. In my case for one of the jobs I had set the
"fixed-delay"
strategy for failure restarts. This led to job being in Terminally failed status . As per the documentation, we went for manual restore with deletion of the FlinkDeployment CRD and restoring from checkpoint. But in our case the application still doesnt start with FlinkDeployment CRD status being in Reconciling state with error as :
Copy code
JobManager deployment is missing and HA data is not available to make stateful upgrades. It is possible that the job has finished or terminally failed, or the configmaps have been del │ │ eted. Manual restore required.
Could someone help with this ?
As a future workaround, we have changed the restart strategy to
exponential-delay
, but still I would like to understand a solution for how to recover from state in case of a terminal failure
a
See https://nightlies.apache.org/flink/flink-docs-release-2.1/docs/deployment/config/#execution-state-recovery-path You can start a job from a given savepoint manually using this. EDIT: never mind. I didn't read you message fully. Seems like you already tried to recover from a checkpoint. I'm not familiar with the issue you describe