Notices by :blank: (i@declin.eu), page 7
-
:blank: (i@declin.eu)'s status on Wednesday, 01-May-2024 03:11:39 JST :blank: @Zergling_man @nozaki and that regs to https://gitlab.com/mangadex-pub/mangadex_at_home are closed -
:blank: (i@declin.eu)'s status on Wednesday, 01-May-2024 03:11:32 JST :blank: @Zergling_man @nozaki i guess not anymore, stopped at 2024 04 01 In conversation from declin.eu permalink Attachments
-
:blank: (i@declin.eu)'s status on Tuesday, 30-Apr-2024 18:22:43 JST :blank: @NEETzsche @mint down is new and down is old networks just don't mix, you're probably better off having people manually mark sections to post with emoji as a quote collector In conversation from social.076.moe permalink -
:blank: (i@declin.eu)'s status on Saturday, 27-Apr-2024 09:13:50 JST :blank: @iamtakingiteasy @mint thanks for the reminder of this, remembered vaguely someone made the :ignore but was confused when i didn't find it anywhere else after In conversation from declin.eu permalink -
:blank: (i@declin.eu)'s status on Friday, 26-Apr-2024 22:15:26 JST :blank: @mint i'd delete mine, but federating three movies worth to random servers is still funny to me In conversation from declin.eu permalink Attachments
-
:blank: (i@declin.eu)'s status on Friday, 26-Apr-2024 21:01:22 JST :blank: welcome @Priceva, i will ban you for doing SEO spam In conversation from declin.eu permalink -
:blank: (i@declin.eu)'s status on Friday, 26-Apr-2024 21:01:19 JST :blank: @dorotthylat too In conversation from declin.eu permalink -
:blank: (i@declin.eu)'s status on Friday, 26-Apr-2024 21:01:15 JST :blank: federating pinned posts for ultimate server reach, gotta get those oembed scrapes up In conversation from declin.eu permalink -
:blank: (i@declin.eu)'s status on Friday, 26-Apr-2024 18:34:17 JST :blank: @verita84eva @colonelj @ChristiJunior @admin @mint lol i think i know why, Create activities with more than 5000 mentions, i bet the indexing blows up too on them
like, https://writeworks.uk/@/the_kyiv_independent/ created 13MB in the DB with just 100 activities
wonder what
select actor, pg_size_pretty(sum(pg_column_size(activities.*))), count(*) from activities group by actor order by sum(pg_column_size(activities.*)) desc limit 100;
says for you, and if there's not a really degenerate software bug out there silently killing ick'oma databaseIn conversation from declin.eu permalink Attachments
-
:blank: (i@declin.eu)'s status on Friday, 26-Apr-2024 06:31:29 JST :blank: @mint @colonelj @verita84eva mint is an alt of everyone else confirmed In conversation from declin.eu permalink -
:blank: (i@declin.eu)'s status on Friday, 26-Apr-2024 06:30:57 JST :blank: @verita84eva @colonelj @mint a week? is someone out there trolling instances they hate somehow, disqordia died when their DB started inflating suspiciously quickly In conversation from declin.eu permalink -
:blank: (i@declin.eu)'s status on Friday, 26-Apr-2024 06:15:53 JST :blank: @mint @colonelj @verita84eva yeah i wonder where the space is spent, a screenshot of https://detroitriotcity.com/phoenix/live_dashboard/ecto_stats?nav=total_table_size could help
but still, the activities tables has no right to be 44GB of millions of CREATE url json blobsIn conversation from declin.eu permalink Attachments
-
:blank: (i@declin.eu)'s status on Friday, 26-Apr-2024 06:13:09 JST :blank: @verita84eva @colonelj yeah that's what the local = 'f' bit means, same as the like/boost pruning in the image
i'll test it on my own install first, or you can go ahead and try in parallel if you trust your own backups tooIn conversation from declin.eu permalink -
:blank: (i@declin.eu)'s status on Friday, 26-Apr-2024 06:11:02 JST :blank: @verita84eva @colonelj damn, the bulk would be in CREATE activities then, doubt much would break if you deleted them, since you're clearing the objects by age too
time to make a dump and report back if anything breaks after deleting a few months of them
pleroma=> select count(*), date_trunc('month', inserted_at) from activities where data->>'type' = 'Create' and local = 'f' group by date_trunc('month', inserted_at) order by date_trunc('month', inserted_at);
count | date_trunc
---------+---------------------
882593 | 2023-03-01 00:00:00
3404857 | 2023-04-01 00:00:00
3695948 | 2023-05-01 00:00:00
3067730 | 2023-06-01 00:00:00
4033487 | 2023-07-01 00:00:00
4050590 | 2023-08-01 00:00:00
604380 | 2023-09-01 00:00:00
19969 | 2023-10-01 00:00:00
712353 | 2023-11-01 00:00:00
1842379 | 2023-12-01 00:00:00
4198836 | 2024-01-01 00:00:00
3653179 | 2024-02-01 00:00:00
4249558 | 2024-03-01 00:00:00
2461312 | 2024-04-01 00:00:00
(14 rows)
pleroma=> select count(*) from activities where data->>'type' = 'Create' and local = 'f' and inserted_at < now() - interval '365 days';
count
---------
3778169
(1 row)In conversation from declin.eu permalink -
:blank: (i@declin.eu)'s status on Friday, 26-Apr-2024 05:33:36 JST :blank: @verita84eva @colonelj if only someone would have been gracious enough to spoon feed you how to clear old activities objects manually In conversation from declin.eu permalink Attachments
-
:blank: (i@declin.eu)'s status on Thursday, 25-Apr-2024 23:51:23 JST :blank: @mint @kirby could also add a config_description for the admin fe scrubs In conversation from declin.eu permalink -
:blank: (i@declin.eu)'s status on Thursday, 25-Apr-2024 23:46:59 JST :blank: @Zergling_man @amerika @p not before the tarballs drop at least In conversation from declin.eu permalink -
:blank: (i@declin.eu)'s status on Thursday, 25-Apr-2024 02:03:52 JST :blank: @p or someone could add a try catch in the follow import loop
https://git.pleroma.social/pleroma/pleroma/-/blob/develop/lib/pleroma/user/import.ex#L43In conversation from declin.eu permalink Attachments
-
:blank: (i@declin.eu)'s status on Thursday, 25-Apr-2024 02:03:45 JST :blank: @p imports keep failing for people midway through their file because oban/DEBUG swallow all the errors, and no one cares to ever fix it on suggestion
too bad https://git.pleroma.social/pleroma/pleroma/-/merge_requests/3589 never went anywhere, even if the foundation of mint is still a bit retarded in other waysIn conversation from declin.eu permalink Attachments
-
:blank: (i@declin.eu)'s status on Thursday, 18-Apr-2024 02:46:49 JST :blank: @7666 @feld people keep getting scammed so fraud departments gave up and started training tellers to just make all big cash withdrawals more difficult
someone's KPI is worth more to them than your freedomIn conversation from declin.eu permalink