Conversation

There's a really big problem with how Mastodon and Pleroma handle servers disappearing for a while.

By default, both daemons will "pause" delivery to remote instances that they haven't seen or been able to deliver to in 7 days. When delivery is paused, even sending DMs to/from test users won't result in a request being sent out.

Delivery is unpaused when you hear from the other server.

See the problem?

If a fediverse server has connectivity issues for a little over 7 days, nobody it used to federate with will send it any messages until they hear from it. And that server won't send any messages to remote servers until it hears from them.

While Mastodon has a way to manually unpause delivery, there is NO indication to the end user that they need to contact an admin to make this happen. If you were to, for example, send a DM to a paused instance after it comes up, you won't get any sort of error saying that delivery is paused, but at the same time, no job will ever be made to deliver your message.

Pleroma does not appear to have a way to unpause delivery at all, meaning that if this happens between two pleroma instances, it's essentially a suspension that neither admin can lift, or even see.

This, by the way, does not fix itself if one of the servers becomes aware of the other again through things like boosts. It *must* be manually cleared by an administrator.

How many instances can you never talk to again, because you or them disappeared for a week?

Anyway, please boost, and maybe someone who works on either project will see this. Eugen actively refuses my reports.