Jobs getting stuck in processing states and can't get cleared in jobs
Expected behaviour
Jobs activated by user should either succeed, have convertion / loading / publishing errors, or fail. If they fail, system admin should be able to successfully retrigger them from the /jobs UI.
Current behaviour
Jobs activated by user are getting stuck in active state and do not clear after retriggering them using /jobs
Steps to reproduce
Go to any book here: https://ncbi.cloud68.co/organizations/8b780e10-b636-42d4-a186-7a4b01f7b9b8
They are stuck in converting. I retriggered via /jobs and they failed on PMC side with wrong files sent to PMC, and are again stuck in Converting on BCMS side.
Seeing same behavior repeatedly - with wrong jobs sent to NCBI and status getting stuck in an active status that won't clear.
NCBI's priority feedback
Y, critical for MVP
QA Steps
- Checked the books on NCBI site which are in status 'Converting' for days -- ALL of them are XML books
- Got some examples from the books stuck on Converting and submitted again. All the books statuses were updated successfully.
- Checked that there are examples of same files on ncbi a) Submitted Oct 11th (other books have been stuck on converting on that day also) Book in status Converting here b) submitted 3 days later book in status previewing
- Check the books stuck in status Converting from /jobs. Checked this book here and clicked the
Retry action
button. To confirm it worked, the last updated dates should change. - All the XML books on the Converting status stuck for days seems to be able to restart conversion via the /jobs retry actions when their statuses appear like this:
- If the column uploaded has an X, the books should be checked as most probably the source files are missing the main_xml tag, and the retry action will not work.