Post APRD6FGo7czGAihQjA by duponin@udongein.xyz
 (DIR) More posts by duponin@udongein.xyz
 (DIR) Post #APPOdmBTTomkzMTbbU by mathias@moc.d-x-b.com
       2022-11-08T23:07:01.169759Z
       
       1 likes, 0 repeats
       
       Finally managed to implement an instance tweak (turned off “busy-waiting” on BEAM) that, so far (knock on wood), doesn’t kill the instance and ALSO lowered CPU usage from a solid 100% to flutter between 7-40% on average.Let’s see if the server is alive tomorrow morning when I wake up. :)
       
 (DIR) Post #APRD6FGo7czGAihQjA by duponin@udongein.xyz
       2022-11-09T20:09:00.572438Z
       
       0 likes, 0 repeats
       
       @mathias do you have the exact settings tweak you made?it interests me
       
 (DIR) Post #APRDxV2oCjmQRVXsLA by mathias@moc.d-x-b.com
       2022-11-09T20:16:54.516390Z
       
       0 likes, 0 repeats
       
       @duponin kind of. I mean, for the disabling of "busy-waiting" for BEAM on #Pleroma I have the suggested settings from the official docs, which is only to be applied on small or few-cores VPS's (which is what I have).https://docs-develop.pleroma.social/backend/configuration/optimizing_beam/As for tweaking postgreSQL I pretty much followed the suggestions on https://pgtune.leopard.in.ua/On top of those things I also created a swap on my server (I normally don't but ran in to a situation where memory got clogged up before being able to write it somewhere so figured I at that point had nothing to lose).I still have more tweaks I need to do if I look at my "Health Check" but I have so far not found any information on how/what/where to fix those things.
       
 (DIR) Post #APRDxVbY7ZZMBFncUS by duponin@udongein.xyz
       2022-11-09T20:18:26.288094Z
       
       0 likes, 0 repeats
       
       @mathias okay, thank you for the links, I'll continue :bunsmile2:
       
 (DIR) Post #APRF2EsFqu45h4QsO8 by mathias@moc.d-x-b.com
       2022-11-09T20:28:58.935659Z
       
       1 likes, 0 repeats
       
       @duponin Just to clarify: it is those *exact* values/settings from those pages that I tweaked. :)The one thing that isn't "obvious" from them is that I increased the number of postgreSQL connections, from the default 100, to 150. The rest was all about specifics of the server CPU/Ram etc. At first I tried to lower the amount of connections. That killed the server completely.If you have any source on admin tweaks for #Pleroma I'm all ears as I need to fill a knowledge gap that I have noticed. :)I also admin an instance of Friendica and it runs...umm...more "evenly" on CPU usage etc, and doesn't generate any errors, which is a thing the Pleroma instance does now and then (which leads me to think I have more tweaking to do).
       
 (DIR) Post #APRF9d6GBBP4r8VWnw by duponin@udongein.xyz
       2022-11-09T20:32:03.956020Z
       
       0 likes, 0 repeats
       
       @mathias I'm not sure having any source but Erlang's documentationI forgot the existence of that Pleroma doc page :blobcatsweats: