Post ASfkbyWhoBLSX1Zlzc by randy@cawfee.club
(DIR) More posts by randy@cawfee.club
(DIR) Post #ASfkbyWhoBLSX1Zlzc by randy@cawfee.club
2023-02-14T10:36:55.843411Z
0 likes, 0 repeats
@alex hey... do we know server stats of spinster or poast? what about horizontal scaling of servers? does pleroma/rebased have means to have load balancers and other stuff so that we don't need to have a single server doing all the things?
(DIR) Post #ASfkbzAPQZ6WWA9TsW by alex@gleasonator.com
2023-02-14T14:06:55.623822Z
1 likes, 1 repeats
@randy Both have hit the upper limit of what Pleroma/Rebased can do. The bottleneck is the database. Load balancing the application server is hard because configuration isn’t synchronized. However Elixir can handle a lot of HTTP requests. The next step would be to try setting up a read-replica in Postgres, which nobody running a big server has tried yet.
(DIR) Post #ASfrTFpUo2BZGk9aRE by feld@bikeshed.party
2023-02-14T15:23:30.009181Z
2 likes, 0 repeats
Nah, Pleroma can do more but we need to refactor. Amusing I just read this right as I was discussing this with Lain again.If we implemented these we'd get way more performancehttps://hexdocs.pm/gen_stage/GenStage.htmlhttps://elixir-broadway.org/