Conversation

Notices

1.  (mint@ryona.agency)'s status on Sunday, 10-Nov-2024 16:54:57 JST 
   Apparently the system crashed nine hours ago, getting stuck at cryptroot-unlock prompt, I was sleeping then. No idea why.
   •  (mint@ryona.agency)'s status on Sunday, 10-Nov-2024 18:40:27 JST 

     in reply to
     Don't know if the crash is related to that, but postgres just started spazzing out after reboot; apparently what's responsible is RUM extension. Switched to default gin indexer, works now. This is also what killed plagu.ee (whch also runs postgres 16 and had no trouble on 15), but I have no idea why this didn't happen here sooner for almost a month.
     Screenshot_20241110_110107.png
     Screenshot_20241110_114400.png
     Screenshot_20241110_114900.png
   •  (mint@ryona.agency)'s status on Sunday, 10-Nov-2024 18:45:30 JST 

     in reply to
     This also means search is broken, but I'd take it over whole instance being broken.
   •  (mint@ryona.agency)'s status on Sunday, 10-Nov-2024 19:34:23 JST 

     in reply to
     Maybe I instead can try one of those fancy search engines pleromer supports, but this'll have to wait until my flesh automaton gets released from his social contract.
   •  (mint@ryona.agency)'s status on Thursday, 21-Nov-2024 04:39:10 JST 

     in reply to
     https://docs.pleroma.social/backend/configuration/search/
     >Note that it's quite a bit more memory hungry than PostgreSQL (around 4-5G for ~1.2 million posts while idle and up to 7G while indexing initially)
     agency=# select count(1) from objects where data->>'type' = 'Note';>18451319
     That's like 15 times the suggested metric and I don't appear to have 64 gigs to spare except on my main workstation that I'd like to use for other purposes.
     † top dog :pedomustdie: likes this.
   •  (mint@ryona.agency)'s status on Thursday, 21-Nov-2024 05:18:20 JST 

     in reply to
     Apparently it also times out because it's using Pleroma.HTTP.get under the hood which respects my proxy settings.
   •  (mint@ryona.agency)'s status on Thursday, 21-Nov-2024 06:33:11 JST 

     in reply to
     Did the retards that wrote the docs ever bother to check them?
     Screenshot_20241121_003242.png
   •  (mint@ryona.agency)'s status on Thursday, 21-Nov-2024 06:35:05 JST 

     in reply to
     For anyone wondering, one has to use just "meilisearch" instead of "search.meilisearch". Either way, got another problem with Jason.DecodeError.
     † top dog :pedomustdie: likes this.
   •  (mint@ryona.agency)'s status on Thursday, 21-Nov-2024 06:43:05 JST 

     in reply to
     Yeah, commited and never tested as usual.
     Screenshot_20241121_003941.png
   •  (mint@ryona.agency)'s status on Thursday, 21-Nov-2024 06:56:42 JST 

     in reply to
     https://github.com/meilisearch/meilisearch/issues/2619
     k, changing meili_post to meili_put in the mix task fixed the issue, now it appears to have started indexing something. Just testing it over SSH tunnel to the workstation now, I wonder if I could use the same instance for all three pleromers I host here, if not that'd be a shame.
     † top dog :pedomustdie: likes this.
   •  (mint@ryona.agency)'s status on Thursday, 21-Nov-2024 07:09:43 JST 

     in reply to
     Felted.
     Screenshot_20241121_010917.png
   •  (mint@ryona.agency)'s status on Thursday, 21-Nov-2024 07:20:52 JST 

     in reply to

     • jae
     @jae I will make a pull request once I'm sure it's working, yeah.
   • jae (jae@fsebugoutzone.org)'s status on Thursday, 21-Nov-2024 07:20:53 JST jae

     in reply to
     @mint
     
     > Yeah, commited and never tested as usual.
     
     the pleroma is plagued? will you cure it?
   •  (mint@ryona.agency)'s status on Thursday, 21-Nov-2024 07:38:46 JST 

     in reply to
     This was as easy as adding a check for nil dates, I guess a couple malformed objects ended up there. Indexed 600k posts so far, system monitor shows around 2,7 gigs used, would probably baloon later.
   •  (mint@ryona.agency)'s status on Thursday, 21-Nov-2024 07:51:18 JST 

     in reply to
     Yup. 1.4m and 5.3 gigs used. Depending on how bad memory usage when idle would be, I could just copy the indexed DB onto the server and be done with it.
   •  (mint@ryona.agency)'s status on Thursday, 21-Nov-2024 08:24:44 JST 

     in reply to

     • jae
     @jae https://git.pleroma.social/pleroma/pleroma/-/merge_requests/4296
   •  (mint@ryona.agency)'s status on Thursday, 21-Nov-2024 09:02:08 JST 

     in reply to
     7.4m, 26.1 gigs, despite that free -h shows less being used. Probably zram kicked in.
   • jae (jae@fsebugoutzone.org)'s status on Thursday, 21-Nov-2024 09:16:06 JST jae

     in reply to
     @mint that's a straight-forward pr, for something that should have likely been straight-forward if i'm reading the diff properly
      likes this.
   •  (mint@ryona.agency)'s status on Thursday, 21-Nov-2024 15:47:46 JST 

     in reply to
     :clueless:
     Phoneposting, will check ram usage later.
     Screenshot_20241121_094610.png
   •  (mint@ryona.agency)'s status on Thursday, 21-Nov-2024 15:52:04 JST 

     in reply to
     Total amount of note objects was about 18 million, 17 seems like a fair amount for those that are public and have at least some text.
     † top dog :pedomustdie: likes this.