Closed kitleyn closed 1 year ago
Thanks for providing logs. This isnt the first time I have had reports of events being added to the database but not being backed up. Thankfully this time there are clear logs showing whats happening. I will investigate and get a fix ASAP
Issue is now patched, fix will be included in the next release
Description
Backup is configured to run from a CloudKey Gen2+ to Backblaze B2. Backup has been working since deployment a couple of weeks ago, including the initial seed of historic events, but over the last week or so I have seen an increasing number of events giving errors which result in a failed upload to Backblaze. These missed events get written to the UPB DB and therefore don't get re-attempted in the regular 'missing event' task. Some events upload without issue. The pertinent error seems to 'no tomes available (503 service_unavailable)' and this is referenced in an article from Backblaze with the suggestion to retry the upload again. I assume that this is built into the rclone functionality, and perhaps not something that UPB is even in control of, but then I noticed that the rclone version (v1.58.1-DEV) included in the docker image is from 2022 (as mentioned in #105) so an upgrade to this would be great in case it is something that has been fixed in a later version. That aside, the issue I am reporting here is that the event entry gets written to the events.sqlite db file despite the upload failing. From looking at the sqlite structure, it doesn't appear to contain any status value so I assume that the event id being present in the DB leads the system to believe it was successfully uploaded to the remote.
The logs show the following
Many thanks