Hi all, in my application, high availability is enabled. One job manager is the leader, and the other is a standby. I have one file that contains 100 records. Now I am sending data to my application. After processing 50 records, I deleted the leader's job manager. After a few seconds, the standby job manager comes up and works as a leader. But I am losing too much data in this process. I expected a 100-record set, but I got only 70 records. Why is HA losing data? Why is standby not taking the whole load of the leader and processing 100 records as expected? please give some suggestions.

Did you have checkpointing enabled? That's a requirement for avoiding data loss.

Yes, checkpointing is enabled in my application and I have given 10s checkpoint interval but still getting data loss. Is there any other way to resolve data loss issue?

What are the source and sink?

Source is custom source where we are reading data from TCP port. Sink is custom sink where we are using GRPC.

Perhaps that explains the problem. In order to avoid data loss, all sources must checkpoint their state, and upon recovery they must be able to rewind back to the position they had at when the checkpoint was taken. If your custom source can not rewind, data will be lost. Similarly, the sink must also participate in checkpointing.

Thanks a lot for your support.