Fixed the issue -- it seems that the default S3 upload timeout of Mastodon is too low for our self-hosted distributed storage cluster, and an underperforming node today triggered a lot of failure. Increased the timeout for now. Note that attachments may take longer to upload until the performance of the cluster restores.
We have updated to Mastodon v3.2.1, and as a change in 3.2.1 introduced stricter authorization requirements in AUTHORIZED_FETCH mode, our federation with Pawoo would probably become buggy and unstable. There's nothing we can do to fix this except urging Pawoo to migrate to a version newer than 3.0.0. We do not want our user content to be fetched by some of our blocked instances.
Administration account for SN.Angry.Im