Hi all, I have a flink job running version 1.10.2, it simply read from a kafka topic with 96 partitions, and output to another kafka topic.
It is running in k8s, with 1 JM (not in HA mode), 12 task managers each has 4 slots. The checkpoint persists the snapshot to azure blob storage, checkpoints interval every 3 seconds, with 10 seconds timeout and minimum pause of 1 second. I observed that the job manager pod memory usage grows over time, any hints on why this is the case? And the memory usage for JM is significantly more compared to no checkpoint enabled. [image: image.png] Thanks a lot! Eleanore