Things should be getting back to normal now.

Something changed somewhere (maybe in the last update and I just never noticed, or maybe the default container ENV changed, not sure), and the result was that all temporary files were trying to be written to an incorrect directory (/opt/mastodon instead of /opt/mastodon/tmp).

The container filesystem is read-only which made those writes fail. They then entered the retry queue which backed up ~1.2m items.

I've triaged and fixed the root cause.