Hi All, I’m running a Flink Session Mode cluster ...
# troubleshooting
a
Hi All, I’m running a Flink Session Mode cluster using the FlinkDeployment CRD and submitting jobs via the REST API. TaskManager pods are created dynamically as jobs are submitted. The issue is that we have over 100 jobs, and when the JobManager pod is restarted, it takes around 20 minutes to recover all the jobs due to the sequential nature of recovery where it start one TM pod at a time and fill all the available slots on that tm pod Is there a better way to use a Flink Session cluster that allows for faster recovery after a JobManager pod restart? Thank You