When syncing job fails it's not scheduled again

[jredrejo]

Observed behavior

A LOD tried to sync and the process didn't finish because the sqlite db was locked by another process (in this case it was sqlitebrowser). That's fine with sqlite.

However when the job failed, kolibri didn't schedule it again, so this LOD stopped syncing until it was restarted.
After restarting it the kolibri.core.auth.tasks.soud_sync_processing job was scheduled again and everything was fine.

Errors and logs

It can be seen beginning on line 550 of this log file
kolibri.txt

Expected behavior

In case a syncing job fails, it should be scheduled again

User-facing consequences

Syncing stop until the kolibri instance of the user is restarted

Steps to reproduce

It happened when the networklocation.sqlite3 file was open in sqlitebrowser but it'll likely happen if db.sqlite3 or job_storage.sqlite3 are open too or kolibri is doing the vacuum task

Context

kolibri 0.16 installed on this PR code #11390

[rtibbles]

Looking at this again - I don't think the database locked error is the issue here, as it is being caught and reported. I think this is actually another instance of #11485, as I am seeing elsewhere in the logs a 404 from the transfersession endpoint.