Akshay Lad

We are launching using Kubernetes & MySQL version is 8.0.33 Both cleanup job & batch task application are scheduled operations. Cleanup job executes every 6 hours whereas task is scheduled...

We use default isolation level. We use hikari datasource & dont override it.

sure will give it a try. Thanks!

Creating new transaction with name [org.springframework.cloud.dataflow.server.service.LauncherInitializationService.initialize]: PROPAGATION_REQUIRED,ISOLATION_DEFAULT I see this line in SCDF server DEBUG logs as well. Does this also need to be updated to ISOLATION_REPEATABLE_READ?

I just checked in our Database, its default isolation level itself is REPEATABLE_READ. 

Hi @cppwfs , We tried adding transaction isolation level = ISOLATION_REPEATABLE_READ in our batch application. But today again got the same deadlock issue.

Hi @cppwfs , We were not able to reproduce issue. However, attaching innodb status file from mysql which shows details about deadlock. [deadlock.txt](https://github.com/user-attachments/files/17375959/deadlock.txt) Hope it helps to investigate further.

DELETE FROM BOOT3_BATCH_JOB_INSTANCE WHERE NOT EXISTS ( SELECT JOB_INSTANCE_ID FROM BOOT3_BATCH_JOB_EXECUTION WHERE BOOT3_BATCH_JOB_INSTANCE.JOB_INSTANCE_ID = JOB_INSTANCE_ID) I believe this query holds the lock of entire table rows which causes the problem.