ableegoldman commented on pull request #8962: URL: https://github.com/apache/kafka/pull/8962#issuecomment-652710895
1. The problem occurs when the original standby task does reach RUNNING and delete the checkpoint, then gets recycled to an active task. If the new active task doesn't get out of CREATED, it won't write the known offsets, but the checkpoint was deleted back when it was a standby. 2. We only pass `stateDirectory.directoryForTaskIsEmpty` in to `initializeOffsetsFromCheckpoint`, not `registerStateStores` -- and even then we only use it to determine whether to throw TaskCorruptedException if we're missing any offsets in the checkpoint ---------------------------------------------------------------- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: us...@infra.apache.org
