Conversation

Notices

1. Embed this notice
   munir (munir@fedi.munir.tokyo)'s status on Sunday, 07-Sep-2025 00:14:25 JST munir
   :02_yell:
   • Embed this notice
     munir (munir@fedi.munir.tokyo)'s status on Sunday, 07-Sep-2025 00:17:42 JST munir

     in reply to

     • 
     • :marseyloadingneon: m0xEE :marseyloading:
     @mint @m0xEE :akko_cry: plz halp?
   • Embed this notice
     munir (munir@fedi.munir.tokyo)'s status on Sunday, 07-Sep-2025 00:55:23 JST munir

     in reply to

     • 
     • :marseyloadingneon: m0xEE :marseyloading:
     @mint @m0xEE yeah apparently its from some old library that pleroma used to use before. deleted the key it was crying about and commands started running (atleast i think so, since they didnt print anything out...), the original reason behind this was pleroma was being sluggish, i checked oban but there werent many jobs open (only about 600?), so i thought that the next thing in checklist would be to analyze and do full vacuum
   • Embed this notice
      (mint@ryona.agency)'s status on Sunday, 07-Sep-2025 00:55:24 JST 

     in reply to

     • :marseyloadingneon: m0xEE :marseyloading:
     @munir @m0xEE Looks like an issue with wrong value in DB config, try nuking it.
   • Embed this notice
     munir (munir@fedi.munir.tokyo)'s status on Sunday, 07-Sep-2025 01:11:16 JST munir

     in reply to

     • 
     • :marseyloadingneon: m0xEE :marseyloading:
     • Phantasm
     @phnt @m0xEE @mint Some ~7GB I assume.
   • Embed this notice
     Phantasm (phnt@fluffytail.org)'s status on Sunday, 07-Sep-2025 01:11:17 JST Phantasm

     in reply to

     • 
     • :marseyloadingneon: m0xEE :marseyloading:
     @munir
     >this is usually due to DB fragmentation
     Post the first 10 entries in this list?
     https://fedi.munir.tokyo/phoenix/live_dashboard/ecto_stats?limit=&nav=bloat&search=&sort_by=waste&sort_dir=desc
     
     And how big is your database?
     
     @mint @m0xEE
   • Embed this notice
      (mint@ryona.agency)'s status on Sunday, 07-Sep-2025 01:11:18 JST 

     in reply to

     • :marseyloadingneon: m0xEE :marseyloading:
     @munir @m0xEE Oban jobs shouldn't be responsible for just the API being slow to respond, this is usually due to DB fragmentation or lack of IOPS.
   • Embed this notice
     munir (munir@fedi.munir.tokyo)'s status on Sunday, 07-Sep-2025 01:15:41 JST munir

     in reply to

     • 
     • :marseyloadingneon: m0xEE :marseyloading:
     • Phantasm
     @phnt @m0xEE @mint I didn't do that, can I do it now?
   • Embed this notice
     Phantasm (phnt@fluffytail.org)'s status on Sunday, 07-Sep-2025 01:15:42 JST Phantasm

     in reply to

     • 
     • :marseyloadingneon: m0xEE :marseyloading:
     @munir @m0xEE @mint DB fragmentation shouldn't be a problem then.
     
     If you used pgTune when setting up the instance, than this is probably lack of IOPS for the DB size you have.
     
     https://pgtune.leopard.in.ua/
   • Embed this notice
     munir (munir@fedi.munir.tokyo)'s status on Sunday, 07-Sep-2025 01:41:06 JST munir

     in reply to

     • 
     • :marseyloadingneon: m0xEE :marseyloading:
     • Phantasm
     • Rock DJ :catcrunk:
     @phnt @m0xEE @dj @mint Thanks so much you guys :), hopefully it's all good now
   • Embed this notice
     Phantasm (phnt@fluffytail.org)'s status on Sunday, 07-Sep-2025 01:41:07 JST Phantasm

     in reply to

     • 
     • :marseyloadingneon: m0xEE :marseyloading:
     • Rock DJ :catcrunk:
     @dj @m0xEE @munir @mint Glussy is a retard, Pleroma isn't configured to use that many and there's not point in doing that anyway. 90% of the time Pleroma here uses 10 connections leaving the other 5(?) unused. 90% of the time Pleroma here uses 2 connections leaving the other 8(?) unused.
     
     It just uses memory that can be used elsewhere for no reason.
   • Embed this notice
     Phantasm (phnt@fluffytail.org)'s status on Sunday, 07-Sep-2025 01:41:08 JST Phantasm

     in reply to

     • 
     • :marseyloadingneon: m0xEE :marseyloading:
     @munir @m0xEE @mint Yeah, make sure to put 20 as the number of connections (or more if you are running also something else that uses the DB on the same box) and don't use the "network storage" option as your storage. It will output few configuration lines, which you should then change in your postgresql.conf file.
     
     Depending on how much free RAM you have on the box, maybe increase or decrease the effective_cache_size setting. And maybe increase the random_page_cost to something like 4. The rest should be fine to copy as is.
     
     EDIT: postgresql restart is probably required for all changes to take effect. Pleroma should ideally be shut down for that, but it will survive even if it is not.
   • Embed this notice
     Rock DJ :catcrunk: (dj@parcero.casa)'s status on Sunday, 07-Sep-2025 01:41:08 JST Rock DJ :catcrunk:

     in reply to

     • 
     • :marseyloadingneon: m0xEE :marseyloading:
     • Phantasm
     @phnt @m0xEE @munir @mint
     Only 20 connections? I put 150 after a post I saw from glussy long ago. :anya: