We are using Kafka connector for some data ingestion jobs. The Kafka connect task is running as a thread in the worker JVM. We noticed that in some cases, the task couldn't be killed/removed correctly and became a orphaned thread. After a new task started, it caused the data duplication.
We tried to increase the task.shutdown.graceful.timeout.ms parameter to a bigger number (like 30000), but this didn't help. We have restarted the worker JVM to fix this issue.
Anyone here encountered similar issue and has any suggestion to it?
与恶龙缠斗过久,自身亦成为恶龙；凝视深渊过久,深渊将回以凝视…
