large portions of the API are timing out without any context or stack trace
[info] POST /api/qvitter/statuses/notifications/read.json
...
[error] #PID<0.1802.0> running Pleroma.Web.Endpoint (cowboy_protocol) terminated Server: verge.info.tm:80 (http) Request: POST /api/qvitter/statuses/notifications/read.json
** (exit) exited in: :gen_server.call(#PID<0.1302.0>, {:checkout, #Reference<0.0.4.16730>, true, 15000}, 5000)
** (EXIT) time out
That's all the log says, and I'm in development mode so it should be pretty verbose. In between the request and the timeout, there are a bunch of debug messages, but they all look like database requests from other parallel tasks, so I dunno how to tell which one is related to the task that's timing out. That timeout also happens with /nodeinfo/2.0.json /api/qvitter/statuses/notifications.json, and /api/statuses/public_and_external_timeline.json, /api/statuses/friends_timeline.json, /api/statuses/mentions.json, /api/statuses/public_timeline.json, /api/qvitter/statuses/user_timeline.json?, and /inbox. So my instance pretty much can't deliver any notifications, serve up any timelines, or add stuff to my inbox.
My instance has been working fine until at least a few days ago, so it had to be a recent update that caused that. (Or me screwing something up.)
Even ignoring this problem, it's also a problem that timeouts don't give some indication of how far the application got in handling that request, before timing out. Sorry, I don't know erlang well enough to generate a stack trace at the point where the timeout exception occurred. Just reporting it.