Conversation

Notices

Embed this notice
parker (parker@poa.st)'s status on Sunday, 18-Sep-2022 11:56:16 JST parker
- Alex Gleason
@alex I'm moving my instance from soapbox to rebased, and it seems to be hanging forever at "drop index if exists activities_create_objects_index"

In conversation Sunday, 18-Sep-2022 11:56:16 JST from poa.st permalink
- Embed this notice
  Alex Gleason (alex@gleasonator.com)'s status on Sunday, 18-Sep-2022 11:56:15 JST Alex Gleason
  in reply to
  
  This is during the migration with mix ecto.migrate?
  
  Check htop. Are you using all your RAM? If not you should ramp up the database settings. In particular edit postgresql.conf and set work_mem to a higher value. Mine is 16MB.
  
  Also if Pleroma is running, you should definitely stop it. I see no reason dropping the index should take a long time unless the database has locks. The thing that takes a long time is creating the new index afterwards.
  
  In conversation Sunday, 18-Sep-2022 11:56:15 JST permalink
- Embed this notice
  Alex Gleason (alex@gleasonator.com)'s status on Sunday, 18-Sep-2022 11:57:27 JST Alex Gleason
  in reply to
  - anime graf mays ?️?
  You can always reindex an index at least.
  
  In conversation Sunday, 18-Sep-2022 11:57:27 JST permalink
- Embed this notice
  anime graf mays ?️? (graf@poa.st)'s status on Sunday, 18-Sep-2022 11:57:28 JST anime graf mays ?️?
  in reply to
  - Alex Gleason
  @parker @alex this took 47 minutes on poast's dedicated database hardware but on a 6t/16gb machine it took like 2 hours for poasts database. it will probably take a long time
  
  In conversation Sunday, 18-Sep-2022 11:57:28 JST permalink
  
  Alex Gleason likes this.
- Embed this notice
  parker (parker@poa.st)'s status on Sunday, 18-Sep-2022 11:57:28 JST parker
  in reply to
  - anime graf mays ?️?
  - Alex Gleason
  @graf @alex Ahh, that's good to know. Hopefully it doesn't fuck up. As I didn't backup anything.
  
  In conversation Sunday, 18-Sep-2022 11:57:28 JST permalink
- Embed this notice
  parker (parker@poa.st)'s status on Sunday, 18-Sep-2022 11:57:50 JST parker
  in reply to
  - Alex Gleason
  @alex It moved onto creating the new index, so I'm hopeful.
  
  In conversation Sunday, 18-Sep-2022 11:57:50 JST permalink
  
  Alex Gleason likes this.
- Embed this notice
  anime graf mays ?️? (graf@poa.st)'s status on Sunday, 18-Sep-2022 11:58:37 JST anime graf mays ?️?
  in reply to
  - Alex Gleason
  - pwm
  @pwm @parker @alex i did this on poast and that was the reason for the 13 hour outage on Monday. we couldn’t spin up poast because it was dying trying to clean up.
  adding config :pleroma, :delete_context_objects, sleep_interval_ms: 3_000 to the your instances prod.secret.exs configuration will help but I believe alex has merged this in to soapbox.exs configuration so it should be default off the shelf
  
  In conversation Sunday, 18-Sep-2022 11:58:37 JST permalink
  
  Alex Gleason likes this.
- Embed this notice
  pwm (pwm@poa.st)'s status on Sunday, 18-Sep-2022 11:58:38 JST pwm
  in reply to
  - anime graf mays ?️?
  - Alex Gleason
  @parker @graf @alex "Fuck it, we're doing it live"
  
  In conversation Sunday, 18-Sep-2022 11:58:38 JST permalink
- Embed this notice
  anime graf mays ?️? (graf@poa.st)'s status on Sunday, 18-Sep-2022 12:05:37 JST anime graf mays ?️?
  in reply to
  - Alex Gleason
  @parker @alex do you mean kB? I don't think I would have set it to that high on a VPS
  
  In conversation Sunday, 18-Sep-2022 12:05:37 JST permalink
  
  Alex Gleason likes this.
- Embed this notice
  parker (parker@poa.st)'s status on Sunday, 18-Sep-2022 12:05:38 JST parker
  in reply to
  - Alex Gleason
  @alex And my work_mem is 3495mb, what graf set it to
  
  In conversation Sunday, 18-Sep-2022 12:05:38 JST permalink
- Embed this notice
  feld (feld@bikeshed.party)'s status on Sunday, 18-Sep-2022 12:06:54 JST feld
  in reply to
  - Alex Gleason
  Lmao that basically means a single worker/connection can consume almost 4GB of ram for a single query
  
  In conversation Sunday, 18-Sep-2022 12:06:54 JST permalink
  
  Alex Gleason likes this.
- Embed this notice
  Alex Gleason (alex@gleasonator.com)'s status on Sunday, 18-Sep-2022 12:09:09 JST Alex Gleason
  in reply to
  - feld
  Lol yeah my man's running Postgres on a supercomputer.
  
  In conversation Sunday, 18-Sep-2022 12:09:09 JST permalink
- Embed this notice
  Salastil (salastil@pleroma.salastil.com)'s status on Sunday, 18-Sep-2022 12:47:23 JST Salastil
  in reply to
  - Alex Gleason
  Why does this need to be done at all? What is the difference between Soapbox BE and Rebased? I’ve been on Soap BE’s dev line the entire time, I’m obviously missing some sort of explainer that was done in an offhand comment over Fedi. I think you should make a blog post because I was under the assumption that I would just git pull one day and various mentions to Pleroma / Soapbox BE would be replaced with Rebased and slowly the projects would diverge over time, the context in this thread makes it seem like there is some sort of labor intensive database migrations that have to be done.
  
  In conversation Sunday, 18-Sep-2022 12:47:23 JST permalink
- Embed this notice
  Alex Gleason (alex@gleasonator.com)'s status on Sunday, 18-Sep-2022 12:49:13 JST Alex Gleason
  in reply to
  - anime graf mays ?️?
  - pwm
  It's probably just an index issue. You can run the `reindexdb` command and name the index from earlier to make it reindex.
  
  In conversation Sunday, 18-Sep-2022 12:49:13 JST permalink
- Embed this notice
  parker (parker@poa.st)'s status on Sunday, 18-Sep-2022 12:49:14 JST parker
  in reply to
  @pwm @graf @alex Yeah, can someone help when they have time? I can turn my instance back on, it works fine, but when I go to my local timeline it crashes.
  
  In conversation Sunday, 18-Sep-2022 12:49:14 JST permalink
- Embed this notice
  parker (parker@poa.st)'s status on Sunday, 18-Sep-2022 12:49:15 JST parker
  in reply to
  @pwm @graf @alex Know what, I'm gonna setup a new vps with buyvm, wanted an excuse to anyways, and try to move my database over.
  
  In conversation Sunday, 18-Sep-2022 12:49:15 JST permalink
- Embed this notice
  parker (parker@poa.st)'s status on Sunday, 18-Sep-2022 12:49:15 JST parker
  in reply to
  @pwm @graf @alex Ah nvm, they won't provision one till Monday.
  
  In conversation Sunday, 18-Sep-2022 12:49:15 JST permalink
- Embed this notice
  parker (parker@poa.st)'s status on Sunday, 18-Sep-2022 12:49:16 JST parker
  in reply to
  @pwm @graf @alex If I accidentally destroy it, I'll be freed from my shitty database and can start over.
  
  In conversation Sunday, 18-Sep-2022 12:49:16 JST permalink
- Embed this notice
  parker (parker@poa.st)'s status on Sunday, 18-Sep-2022 12:49:16 JST parker
  in reply to
  @pwm @graf @alex Eh, looks like I broke it anyways
  
  In conversation Sunday, 18-Sep-2022 12:49:16 JST permalink
- Embed this notice
  Alex Gleason (alex@gleasonator.com)'s status on Sunday, 18-Sep-2022 12:50:02 JST Alex Gleason
  in reply to
  You should take this opportunity to pg_dump -Fc pleroma > pleroma.dump
  
  In conversation Sunday, 18-Sep-2022 12:50:02 JST permalink
- Embed this notice
  anime graf mays ?️? (graf@poa.st)'s status on Sunday, 18-Sep-2022 12:56:04 JST anime graf mays ?️?
  in reply to
  - Alex Gleason
  - pwm
  @alex @parker @pwm everybody should be running this on a cronjob weekly at the very least. with a custom backup even poasts database backups only take 17GB tar'd
  
  In conversation Sunday, 18-Sep-2022 12:56:04 JST permalink
  
  Alex Gleason likes this.

Public

Conversation

Notices

Feeds