Half the time I can't even post on my own website without copying what I wanted to post to a duplicate tab and trying again. It pisses me off immensely, but I don't even know where to start looking to fix this. More humiliating than my sins, and that says a lot.
@grips bloatfe doesn't have this issue (no soyscript)
@romin Goes through after about five seconds. Suddenly it's hard to reproduce. 
@grips so does it work or what
@romin Suddenly, yeah. It most often fails with media, but now it's working somehow.
@grips @romin @romin IO pressure makes Postgres queries fall over. Seen it many times over here when autovacuum hits a larger table or Gitea decides to do its thing. Same happens with favs and repeats that don't appear to go through.
The reason why it likely hits more with media is because the files are still in a write cache/disk backlog when sending the post.
The reason why it likely hits more with media is because the files are still in a write cache/disk backlog when sending the post.
@grips @romin @romin I don't remember this happening when I was using BuyVM's slabs, but I also had lvmcache in front of it that resided on the main disk. Still notifs wouldn't load half the time after a refresh.
I don't know how Postgres stores tables, but if it stores them in one giant file and that file is on the slab, it's entirely possible that slabs would be the issue. It could also be that CPU cores are stuck in iowait stalling everything else on the system.
I don't know how Postgres stores tables, but if it stores them in one giant file and that file is on the slab, it's entirely possible that slabs would be the issue. It could also be that CPU cores are stuck in iowait stalling everything else on the system.
Follow
@phnt @grips @romin@cawfee.club @romin@shitposter.world big tables get split among multiple files, default is 1GB. indexes are stored separate i believe
