Conversation

Notices

1. Embed this notice
   Alex Gleason (alex@gleasonator.com)'s status on Saturday, 15-Oct-2022 03:20:32 JST Alex Gleason

   • ??? ???? ?????? :verified:
   Sorry, the table is called data_migrations
   • Embed this notice
     ??? ???? ?????? :verified: (ryan@ryanthomas.org)'s status on Saturday, 15-Oct-2022 06:16:30 JST ??? ???? ?????? :verified:

     in reply to

     • ??? ???? ?????? :verified:

     I took a nap so left it alone for a while, bumped remote_post_retention_days down to 28 (not that it matters here, I had meant to do that anyway), restarted pleroma, then ./bin/pleroma_ctl database vacuum analyze and rebooted when done. So far, memory is still higher than I remember it being in the past, which you’ve already explained, but swap is lower, and things that weren’t loading are loading fine, and I haven’t seen any more gateway timeout errors.

     If I get issues again, could it be helpful to purge old remote content and do a vacuum full? Is it a good idea to do that periodically anyway? And should I stop my pleroma service while doing so or just expect it may cause a few hours downtime?

     Appreciate your help and anyone reading this who might be inspired to chime in! Best

     Alex Gleason likes this.
   • Embed this notice
     ??? ???? ?????? :verified: (ryan@ryanthomas.org)'s status on Saturday, 15-Oct-2022 06:16:31 JST ??? ???? ?????? :verified:

     in reply to

     Anything of note here? Only two output rows from that sql command:

     1 | populate_hashtags_table | 3 | f | {} | {"state": "complete", "message": null, "fault_rate": 0.0, "started_at": "2022-08-31T18:21:20.383740", "iteration_processed_count": 0} | 2022-08-31 17:59:53 | 2022-08-31 17:59:532 | delete_context_objects | 3 | f | {} | {"state": "complete", "message": null, "fault_rate": 0.0, "started_at": "2022-10-13T22:43:16.525248", "failed_count": 0, "affected_count": 272949, "processed_count": 272949, "max_processed_id": 973105, "records_per_second": 31.07696686781282, "iteration_processed_count": 272949} | 2022-10-13 22:40:46 | 2022-10-13 22:40:46
   • Embed this notice
     Alex Gleason (alex@gleasonator.com)'s status on Saturday, 15-Oct-2022 06:19:20 JST Alex Gleason

     in reply to

     • ??? ???? ?????? :verified:
     Vacuuming could definitely help, as well as reindexing, or even the nuclear "dump and reimport" option. A lot of problems I've noticed lately seem to be related to indexes not working.
     
     Also, possibly relevant: https://gitlab.com/soapbox-pub/rebased/-/issues/111
   • Embed this notice
     anime graf mays ?️? (graf@poa.st)'s status on Saturday, 15-Oct-2022 06:26:45 JST anime graf mays ?️?

     in reply to

     • ??? ???? ?????? :verified:
     @alex @ryan for what tis worth we didnt dump anything. we just used pg_repack and the difference was night and day. I used to have it on a schedule now it's back on a schedule every sunday night
     Alex Gleason likes this.