Dear Flink team, Here is my case. Assistance on priority will be greatly appreciated. As part of EMR to EKS migration, We deployed

Flink-1.15

in EKS and testing where its been observed the pod disk usage growing on subsequent deploys(like upgrademode laststate/ suspend, before checkpoint completes or even after checkpoint completes). We found the taskmanager local state being retained and not getting cleaned up under

/data

. And hence the disk usage is doubled. I would like to understand what folders will be retained on subsequent deploys. Below sample data being retained in all subsequent deploys.

1001M	2023-03-08 17:58	/data/flink-local-data/tm_flink-personas-experiment-taskmanager-1-2/localState/aid_6b6fd573128d17d190e0b4fdc33e6387/jid_f92a1778225e260ac9febc8dea7aba5e/vtx_f7c20420360e0172e54933e6abdd481a_sti_1/chk_475590/b603b845d9f14110a6d8b32562e04fe0
1001M	2023-03-08 17:58	/data/flink-local-data/tm_flink-personas-experiment-taskmanager-1-2/localState/aid_6b6fd573128d17d190e0b4fdc33e6387/jid_f92a1778225e260ac9febc8dea7aba5e/vtx_f7c20420360e0172e54933e6abdd481a_sti_1/chk_475590
1001M	2023-03-08 17:58	/data/flink-local-data/tm_flink-personas-experiment-taskmanager-1-2/localState/aid_6b6fd573128d17d190e0b4fdc33e6387/jid_f92a1778225e260ac9febc8dea7aba5e/vtx_f7c20420360e0172e54933e6abdd481a_sti_1

Assistance on priority will be greatly appreciated.

Please keep in mind that this is an open-source, non-commercial channel. There is no support SLA from the open source community