The index status of document/chunk is always running after celery worker restart

[Mini256]

Problem

During the index building process, the celery worker restart may result in an inconsistency between the celery task status in redis and the index status in the tidb, and the index status of the document/chunk is always in the running state.

Solution

Option 1

Using celery task status as the source of index build status

Option 2

Update the index status of uncompleted tasks to the correct status after the celery worker restarts