I’ve got two different installations where backup configurations that were previously working are now failing. Backups are being created locally, but are not being uploaded to AWS S3.
File uploads are still working. There’s been no change in the configuration at AWS or in discourse. I’m thinking that maybe something has changed in discourse?
I’ve gone over all the configuration in discourse and on S3, and it all looks like correct to me.
I’ve looked over the unicorn logs, and I see the backup job going through without apparent error, but also with no mention of the S3 upload.
One of the installations involved had a disk crash, and I recovered it from backup. The S3 backups had stopped working a week before (though fortunately I was able to find the later backup file in the mess of un-named files that fsck was able to recover from the old disk). They still don’t work after recovery of that site from backup, but the fact that I have another installation on a different server also failing suggests to me that it’s a broader problem.
Apr 1 and May 3 are the last dates of the backups uploaded from each of the two servers. Both servers are currently running Discourse v1.8.0.beta12 +20.
Any idea what might be going wrong, or how I can further diagnose this?