Mastodon shown as running, but won't load

[scooke]

The app dashboard indicate everything is fine. But all that loads is an error page stating that there was something wrong on the server end, or that a particular page doesn't exist. There is enough ram for app and redis. For awhile initially, when I first noticed a problem, there was a pop up error about Postregsql not being able to be used because something else was using it.

Any ideas?

[girish]

@scooke Does restarting the app help? Anything in the logs? Maybe you can restart postgresql as well?

[scooke]

@girish I've bumped the ram for the app, redis and postgresql, restarted the app, and postgrsql. Still nothing. Shows as running, but it won't load in my web (desktop nor android). Logs look like this generally:

Aug 18 16:33:09 172.18.0.1 - - [18/Aug/2020:14:33:09 +0000] "POST /inbox HTTP/1.1" 202 31 "-" "http.rb/4.3.0 (Mastodon/3.1.2; +https://mamot.fr/)"
Aug 18 16:33:09 Rejected Create activity https://mamot.fr/users/Anchor/statuses/104708168537508500/activity from https://mamot.fr/users/Anchor
Aug 18 16:33:09 2020-08-18T14:33:09.990Z pid=21 tid=gogyi5gcp class=ActivityPub::ProcessingWorker jid=4e80c954c6e2d2ad87992b0d elapsed=0.031 INFO: done
Aug 18 16:33:10 [ee9b503a-0e1a-47e0-ae57-7fbecae91e60] method=GET path=/about format=html controller=AboutController action=show status=301 duration=1.15 view=0.00 db=0.00 location=https://my.mastodon.com/about
Aug 18 16:33:10 172.18.0.1 - - [18/Aug/2020:14:33:10 +0000] "GET /about HTTP/1.1" 301 120 "-" "Mozilla (CloudronHealth)"
Aug 18 16:33:20 [2ea7da8a-abbd-46aa-8d45-34803ce2a8ce] method=GET path=/about format=html controller=AboutController action=show status=301 duration=1.16 view=0.00 db=0.00 location=https://my.mastodon.com/about
Aug 18 16:33:20 172.18.0.1 - - [18/Aug/2020:14:33:20 +0000] "GET /about HTTP/1.1" 301 120 "-" "Mozilla (CloudronHealth)"
Aug 18 16:33:25 2020-08-18T14:33:25.234Z pid=21 tid=gogyi67hx class=ActivityPub::ProcessingWorker jid=f4210b6ff5c2ff31e26ffec8 INFO: start
Aug 18 16:33:25 [a0ab4d86-41d6-4865-b50f-b539f641befb] method=POST path=/inbox format=html controller=ActivityPub::InboxesController action=create status=202 duration=22.80 view=0.00 db=2.94 key=https://mamot.fr/users/Anchor#main-key

I'm not sure what that mamot.fr url is.

[nebulon]

Are you able to restore the app to the latest backup and see if that fixes it? Maybe something went wrong during postgres update

[girish]

Actually, I would check if it's some DNS issue. Did you try another machine/phone, just to check if it's not some DNS error? Because the app seems to be responding.

[scooke]

@girish Hm, I am using the Namecheap API for that domain. Maybe I will revert back to Wildcard.

I've reverted to Wildcard, and it went through, but the app is still not reachable, though running. I'll try the whole restart thingy again.

I did notice other user names appearing in the logs... why would they be accessed? Like, https://mamot.fr/users/Anchor.

[scooke]

@scooke I realize this is for a different app, but a similar error has now popped up with my Matrix server, here is the Postgresql error:

Error : - Error clearing postgresql. Status code: 500 message: database "longstringofcharacters" is being accessed by other users

This is what I saw on the Mastodon dashboard when I initially investigated why it wasn't running.

[necrevistonnezr]

Is it just me or is there a general problem with the new Postgres version?

[scooke]

@necrevistonnezr I did manage to get Matrix running again after hitting the Retry button on the Task Error pane. But my mastodon instance is just dead and I'm going to just reinstall totally. What a pain.

[girish]

@necrevistonnezr said in Mastodon shown as running, but won't load:

Is it just me or is there a general problem with the new Postgres version?

We have heard of some issues with postgres but so far it's been related to migration issues and not the database itself. We are fixing the migration code so this doesn't happen again.

What happens during migration is: we export the old database, run new postgres, reimport the database. The re-import takes a lot of memory and it's dependent on how much data you have in postgres. The fix we are making now is to give the database more memory when re-importing and then scaling it down after the import is over.

If you hit the above, the fix is easy, just give postgres more memory and restart the code with systemctl restart box. It will them re-import (it keeps track that it failed the last time around).