Post AxAq3q4mulFM8RHboe by pernia@cum.salon
(DIR) More posts by pernia@cum.salon
(DIR) Post #AxApzpEQ7404MCsHlA by pernia@cum.salon
2025-08-14T22:06:26.318849Z
1 likes, 1 repeats
@lanodan @lain hey, how long should a db prune (using pleroma commands) take? i have a 66gb db on a 4 core arm server with 24gb of ram (with postgres optimized for it), and the prune i did took 17 days before i gave up and stopped it.
(DIR) Post #AxAq3q4mulFM8RHboe by pernia@cum.salon
2025-08-14T22:07:10.716042Z
0 likes, 1 repeats
@lain @lanodan i used this command: MIX_ENV=prod mix pleroma.database prune_objects --keep-threads --prune-orphaned- activities --vacuum
(DIR) Post #AxAqsigQA3OvyV9CMK by pedophilesoftwareinc@cum.salon
2025-08-14T22:16:19.686128Z
1 likes, 1 repeats
@boyperpia @lain @lanodan again, akkoma's --keep-threads was contributed without any actual testing on whether it would work ok on big ass databases, it's probably missing more than a couple of indexes for it to actually function
(DIR) Post #AxAr3yUrhstgCn6pFo by churkia@cum.salon
2025-08-14T22:18:23.506944Z
3 likes, 1 repeats
@boyperpia @lain @lanodan i would not be surprised in the least if dumping the objects table as csv and awking it would be faster
(DIR) Post #AxArIyjv5PWrInIyps by pedophilesoftwareinc@cum.salon
2025-08-14T22:21:08.402405Z
2 likes, 1 repeats
@churkia @lain @lanodan @p apotheosis this @boyperpia ggs ass hole already
(DIR) Post #AxBIjfoKGkRHAzJsMS by pernia@cum.salon
2025-08-15T03:28:28.940674Z
0 likes, 1 repeats
@pedophilesoftwareinc @lain @lanodan you think i should try it without --keep-threads?
(DIR) Post #AxBJXECJCyx40yAPFA by pernia@cum.salon
2025-08-15T03:37:25.502696Z
0 likes, 1 repeats
@pedophilesoftwareinc @lain @lanodan also, It seems 45.148.18.72 has been scraping the shit out of the salon, and also added "smega" to the remoteIP field in adminfe (which makes pleroma crash after trying to read it, due to a type mismatch of some sort)
(DIR) Post #AxC2rbx55zVTPt8bdg by lain@lain.com
2025-08-15T12:05:20.522286Z
3 likes, 1 repeats
@pernia @lanodan i don't recommend running this at all. i'd remove it altogether but other maintainers like it.
(DIR) Post #AxC2w5juHw9QxX70hU by hj@shigusegubu.club
2025-08-15T12:06:08.508093Z
1 likes, 0 repeats
@lain @pernia @lanodan we should have serious discussion about it :assburger:
(DIR) Post #AxC33VrxXJus1iPw7E by piggo@piggo.space
2025-08-15T12:06:28.887788Z
2 likes, 0 repeats
@lain @pernia @lanodan the only safe way to maintain pleroma is to keep buying more disk
(DIR) Post #AxC33WjUKIjahjSgYC by lain@lain.com
2025-08-15T12:07:28.520252Z
1 likes, 1 repeats
@piggo @pernia @lanodan no, you can delete old activities and objects, it's safe.
(DIR) Post #AxC34MXDj9opMk7ceG by lanodan@queer.hacktivis.me
2025-08-15T12:07:33.208358Z
1 likes, 0 repeats
@lain @pernia > other maintainers like it.Well I don't think I ever used it myself…
(DIR) Post #AxC3ReMVc1FZJgDSQi by hj@shigusegubu.club
2025-08-15T12:11:50.818276Z
1 likes, 0 repeats
@lain @pernia @piggo @lanodan it is somewhat problematic if someone reprööts deleted activities but I don't remember exactly what was happening.
(DIR) Post #AxC3kZpofWfYmFwvSq by lanodan@queer.hacktivis.me
2025-08-15T12:15:11.265932Z
2 likes, 0 repeats
@hj @pernia @lain @piggo Pretty sure it'll just refetch that deleted activity?
(DIR) Post #AxC3pZQLciUex2Pxo0 by hj@shigusegubu.club
2025-08-15T12:16:09.040304Z
1 likes, 0 repeats
@lanodan @pernia @lain @piggo i'm not sure. It hasn't happened since. Maybe i should prune db just so that it happens again.
(DIR) Post #AxC4AjzpVkRXp16pGK by lucy@pleromertest.cirno.systems
2025-08-15T12:14:27.234390Z
1 likes, 0 repeats
@lain @pernia @lanodan why is pruning a bad idea?
(DIR) Post #AxC4Al33bEcR5Jcuvo by lain@lain.com
2025-08-15T12:19:57.143506Z
2 likes, 1 repeats
@lucy @pernia @lanodan it just removes so little and is so heavy on the db. it's much easier and faster to just delete old activities. i think we should set other expectations for database size. soykaf is at ~300gb and i don't think that's excessive for a 7 year old instance. it's also not expensive to host.
(DIR) Post #AxC4Tr7bF9lTQGrGOO by lucy@pleromertest.cirno.systems
2025-08-15T12:23:03.210574Z
1 likes, 0 repeats
@lain @pernia @lanodan im at 175MB on this instance which is mostly down unless i work on pleromer.is this 300GB while deleting old remote posts or the full backlog of time?
(DIR) Post #AxC4TrpCd2dvbVG5M8 by lain@lain.com
2025-08-15T12:23:26.378262Z
1 likes, 1 repeats
@lucy @pernia @lanodan nothing deleted, it's the full db of the very first pleroma instance
(DIR) Post #AxC4YUUumnwehgyR4y by lucy@pleromertest.cirno.systems
2025-08-15T12:23:59.924138Z
3 likes, 0 repeats
@lain @pernia @lanodan i imagined it'd be bigger
(DIR) Post #AxC4YVeAVswQGgJL8q by lain@lain.com
2025-08-15T12:24:16.822177Z
3 likes, 3 repeats
@lucy @pernia @lanodan hey
(DIR) Post #AxC4rBAmrJVCGtTE7E by lanodan@queer.hacktivis.me
2025-08-15T12:27:34.829151Z
2 likes, 0 repeats
@lucy @pernia @lain Well soykaf is also not a big instance so probably doesn't receives everything including hellthreads.(Imagine the size of FSE…)
(DIR) Post #AxC6HOAtF2AW4wa240 by phnt@fluffytail.org
2025-08-15T12:43:36.387368Z
1 likes, 1 repeats
@pernia @lain @lanodan cawfee took a month I think and in the end grips ran it on his laptop for two days after giving up on doing it on the BuyVM VM.
(DIR) Post #AxC6IARBnjj9lOZj8K by phnt@fluffytail.org
2025-08-15T12:43:46.877865Z
3 likes, 2 repeats
@lucy @pernia @lain @lanodan Consult the image.pleroma-pruned-db-winning.png
(DIR) Post #AxC6jpoC5o9TtNWoam by phnt@fluffytail.org
2025-08-15T12:48:44.637501Z
2 likes, 1 repeats
@lain @pernia @lucy @lanodan I would prefer a delete Announce activities from remote relays mix task. That frees up few GB every ~6 months and is totally useless data otherwise. Or drop the activities upon receiving the Announce if the fetch was successful and that would also solve the ghost repeats on posts.
(DIR) Post #AxC7JvWsIi4w3kdaJE by snacks@netzsphaere.xyz
2025-08-15T12:55:14.974634Z
1 likes, 0 repeats
@pernia @lain @lanodan wtf, last time i pruned at about 70gb it took less than a day on a single core shitbox with the db on networked storage iirc
(DIR) Post #AxC7Yn2GA9QS3nP9ay by phnt@fluffytail.org
2025-08-15T12:57:56.902053Z
4 likes, 1 repeats
@snacks @pernia @lain @lanodan I/O in VMs is notoriously shit since they don't want you do host DBs on there. It literally takes 4 hours here to repack the activities and objects tables (~30GB), but I get twice as much IOPS compared to BuyVMs slab storage with lvmcache.
(DIR) Post #AxC7hnEGoOaRGdJcmm by snacks@netzsphaere.xyz
2025-08-15T12:59:32.940407Z
1 likes, 0 repeats
@phnt @pernia @lain @lanodan the networked storage on hetzner was so slow it caused constant errors, idgi
(DIR) Post #AxC7xsqhgWjJKXxtbs by phnt@fluffytail.org
2025-08-15T13:02:29.322115Z
3 likes, 1 repeats
@snacks @pernia @lain @lanodan You might had higher throughput and instead was limited by IOPS. The BuyVM slab can 200 IOPS max, but sequential transfer speeds were around 250 MB/s. Here, I can do 500 IOPS both ways, but I am limited to 35MB/s in reads and writes separately. On paper it looks like a downgrade, but Postgres performance is almost incomparable.
(DIR) Post #AxCAYaRNI2u4rOkMTo by piggo@piggo.space
2025-08-15T13:16:16.811528Z
2 likes, 0 repeats
@hj @pernia @lain @lanodan if you do this, be extra careful when pruning old uploads using find. I deleted all my files by having the arguments in a wrong order
(DIR) Post #AxCAcRKv54kXmaZ56G by hj@shigusegubu.club
2025-08-15T13:32:14.008437Z
3 likes, 0 repeats
@piggo @pernia @lain @lanodan >pruning old uploadsnever :ARUWaiting2:
(DIR) Post #AxCJfyimJRIXmQUEK0 by pernia@cum.salon
2025-08-15T15:13:43.137015Z
0 likes, 1 repeats
@lain @piggo @lanodan could u give me some pointers? I like the ideas of keeping threads and such like the commands do
(DIR) Post #AxCJotcscqAAaGypP6 by pernia@cum.salon
2025-08-15T15:15:20.236143Z
0 likes, 1 repeats
@phnt @lain @lanodan ill try it on my beefy desktop