Hi Abhinav If the leader jobmanager fails during savepoint, that savepoint would fail and new jobmanager would then restore from previous jobgraph with latest completed checkpoint in the high-availability storage. That's why new jobmanager could not know anything with regard to previous savepoint.
Best Yun Tang ________________________________ From: Bajaj, Abhinav <abhinav.ba...@here.com> Sent: Saturday, July 27, 2019 7:25 To: user@flink.apache.org <user@flink.apache.org> Subject: Savepoint process recovery in Jobmanager HA setup Hi, I am trying to test a scenario that triggers a savepoint on a Flink 1.7.1 Job deployed with jobmanager HA mode. The purpose is to check if savepoint process recovers if the leader jobmanager fails during the savepoint. During my testing, I found that the new leader jobmanager returns the below error for the savepoint trigger request – {"errors":["Operation not found under key: org.apache.flink.runtime.rest.handler.job.AsynchronousJobOperationKey@e287af3"]} Does Flink support savepoint process recovery in Jobmanager HA setup? If yes, can you please suggest how to find the savepoint request? Appreciate your time and help. ~ Abhinav Bajaj