Post Azees2hC2TfRQDsmwK by john_darksoul@ariamispainted.world
 (DIR) More posts by john_darksoul@ariamispainted.world
 (DIR) Post #AzedaI6jtmry3Snrii by john_darksoul@pl.firelink.digital
       2025-10-28T02:52:55.997182Z
       
       0 likes, 0 repeats
       
       It’s 100% a database issue. I can post and I can see the posts from here but posts vanish the second I refresh
       
 (DIR) Post #AzedeNpqXuHMycqxea by john_darksoul@pl.firelink.digital
       2025-10-28T02:53:40.363930Z
       
       0 likes, 0 repeats
       
       So what do I do here? My front end is effectively disconnected from my database? Does anyone have any idea?
       
 (DIR) Post #AzeeCieRetApNmtPxg by WandererUber@poa.st
       2025-10-28T02:59:53.351678Z
       
       1 likes, 0 repeats
       
       @john_darksoul Test connection manually: psql -U akkoma -h localhost -d akkoma (enter password). If it connects, run \dt to list tables (should see activities, objects for posts).
       
 (DIR) Post #AzeeI9iKT3WIlQImIq by skylar@misskey.yandere.love
       2025-10-28T03:00:50.631581Z
       
       1 likes, 0 repeats
       
       @john_darksoul try it on another deviceif it works, it's the cache on the device(guess what my pleroma did once)
       
 (DIR) Post #AzeeOTgR8OvSR4QnmC by john_darksoul@pl.firelink.digital
       2025-10-28T03:02:00.226342Z
       
       0 likes, 0 repeats
       
       @WandererUber I think it shipped with enable media proxy. I don’t know why this would cause an issue (I don’t know what this does) but disabling fixed it.
       
 (DIR) Post #AzeeenAWSMi5V9HbX6 by WandererUber@poa.st
       2025-10-28T03:04:57.975097Z
       
       0 likes, 0 repeats
       
       @john_darksoul It's a built-in reverse proxy that fetches and caches remote media Media proxy uses Oban (Akkoma’s job queue) to fetch and cache media in the background. If it fails repeatedly (e.g. timeout, DNS, SSL), it can clog the queue. If hundreds of media fetch jobs fail, Oban gets overwhelmed → delays or fails other critical jobs like inserting your post into the database or delivering it to followers.Posts appear to "send" (UI says OK), but the database transaction fails silently due to Oban/Repo contention → post never persists → vanishes on refresh.
       
 (DIR) Post #Azees1lPVJRkX0qdsG by john_darksoul@ariamispainted.world
       2025-10-28T03:06:02.420782Z
       
       0 likes, 0 repeats
       
       @WandererUber @john_darksoul Why would notifications and timeline entries affect this?
       
 (DIR) Post #Azees2hC2TfRQDsmwK by john_darksoul@ariamispainted.world
       2025-10-28T03:06:42.170793Z
       
       0 likes, 0 repeats
       
       @WandererUber @john_darksoul Sorry, other way around. Why would it affect them?
       
 (DIR) Post #Azees3VX1JvvwLQzOy by WandererUber@poa.st
       2025-10-28T03:07:20.043434Z
       
       0 likes, 0 repeats
       
       @john_darksoul @john_darksoul You're asking why would a clogged Oban queue affect notifications and timeline entries?
       
 (DIR) Post #AzefCiZO3tAJ5G5el6 by john_darksoul@ariamispainted.world
       2025-10-28T03:08:12.980267Z
       
       0 likes, 0 repeats
       
       @WandererUber @john_darksoul Why do either them need media proxy? Wouldn't media proxy only affect media? Like uploaded images.
       
 (DIR) Post #AzefCjsvAkNvA8ElqS by WandererUber@poa.st
       2025-10-28T03:11:04.679508Z
       
       0 likes, 0 repeats
       
       @john_darksoul @john_darksoul Media proxy does not affect notifications or timeline entries directly. It only handles remote media. The reason they break when media proxy fails is because failed media jobs clog the Oban job queue, which delays or blocks all background tasks — including saving your posts and sending notifications.
       
 (DIR) Post #AzefQJ67YxeQ8bsggC by john_darksoul@ariamispainted.world
       2025-10-28T03:12:23.454987Z
       
       1 likes, 0 repeats
       
       @WandererUber @john_darksoul Oh ok then I'm back on my shoot the devs train of thought because it came like that.
       
 (DIR) Post #AzefQKXoBVOOclqJTU by WandererUber@poa.st
       2025-10-28T03:13:32.410705Z
       
       0 likes, 0 repeats
       
       @john_darksoul @john_darksoul In ur case it's probably not directly clogging but incorrect setup but I found all this out by literally asking Grok sorry to lead you around by the nose like thatI have absolutely no clue about Akkoma
       
 (DIR) Post #AzefhE01lt4fFU68Js by john_darksoul@ariamispainted.world
       2025-10-28T03:15:32.405745Z
       
       0 likes, 0 repeats
       
       @WandererUber @john_darksoul It's fine. I do think I screwed up setup at some point. I mean for some reason it created postgres as a user and assigned IT as the owner of the db instead of akkoma (the user created specifically to own everything) and I couldn't generate a new password reset token until I manually make akkoma the new owner
       
 (DIR) Post #AzefhF1TxxpeQHmoE4 by Will2Power@poa.st
       2025-10-28T03:16:35.952769Z
       
       0 likes, 0 repeats
       
       @john_darksoul @WandererUber @john_darksoul Doesn't Mister Toe use Akkoma?
       
 (DIR) Post #Azefif68VNqJaCdlhI by WandererUber@poa.st
       2025-10-28T03:16:52.152207Z
       
       0 likes, 0 repeats
       
       @Will2Power @john_darksoul @john_darksoul Grok has no idea
       
 (DIR) Post #AzefpVA1qktEn3irdA by john_darksoul@ariamispainted.world
       2025-10-28T03:17:57.692321Z
       
       1 likes, 0 repeats
       
       @Will2Power@WandererUber Yea, his and radiofreegreenland are akkoma.
       
 (DIR) Post #AzefzxvFllhAtAnjCi by WandererUber@poa.st
       2025-10-28T03:19:58.785884Z
       
       0 likes, 0 repeats
       
       @john_darksoul @john_darksoul double check the setup script if you want to know if it's your fault I guess. Maybe you ran createdb command as the wrong user