Fixing transcription crashes caused by job stoppage
Apparently, scheduling a new job when one is already running (even using the enqueue api) causes the running job to be stopped. We weren't handling that case correctly. This cl makes sure no more work is attempted after a job is stopped by cancelling any
active transcription task. We request that stopped task be rescheduled by the job scheduler, so it will get run eventually.
I was able to verify this fix by sending a new voicemail while backfill old transcription tasks were running.
Bug: 64908823,63524274,65129734,63803709
Test: manual and unit tests
PiperOrigin-RevId: 167617191
Change-Id: Icc92997c2687e61bef9b3b7f9ff572da2cb4ed2e
4 files changed