Conversation
Notices
-
† top dog :pedomustdie: (dcc@annihilation.social)'s status on Tuesday, 25-Feb-2025 17:02:49 JST † top dog :pedomustdie:
Having a high core server is much nicer than you think it is. - homosexual likes this.
-
† top dog :pedomustdie: (dcc@annihilation.social)'s status on Tuesday, 25-Feb-2025 17:05:51 JST † top dog :pedomustdie:
@tyler 20 cores 40 threads -
Tyler (tyler@nicecrew.digital)'s status on Tuesday, 25-Feb-2025 17:05:52 JST Tyler
How many cores? -
† top dog :pedomustdie: (dcc@annihilation.social)'s status on Tuesday, 25-Feb-2025 17:06:30 JST † top dog :pedomustdie:
@tyler also i want to figure out how to load like only a part of a db in ram to make it faster -
Tyler (tyler@nicecrew.digital)'s status on Tuesday, 25-Feb-2025 17:08:03 JST Tyler
Pretty cool. Once I got 12/24 in my workstation I was glad. Maybe in another couple years I'll upgrade to the ddr5/am 5 socket. I didn't know it existed when I built this one. † top dog :pedomustdie: likes this. -
Tyler (tyler@nicecrew.digital)'s status on Tuesday, 25-Feb-2025 17:13:42 JST Tyler
https://www.postgresql.org/docs/current/sql-createtable.html
Unlogged temporary tables will be kept in ram and are extremely fast, but not crash safe.† top dog :pedomustdie: likes this. -
† top dog :pedomustdie: (dcc@annihilation.social)'s status on Tuesday, 25-Feb-2025 17:14:44 JST † top dog :pedomustdie:
@tyler I guess i am using them rn (i am repacking the db) In conversation permalink -
Tyler (tyler@nicecrew.digital)'s status on Tuesday, 25-Feb-2025 17:16:34 JST Tyler
Good luck nigga hope you don't crash In conversation permalink † top dog :pedomustdie: likes this. -
Tyler (tyler@nicecrew.digital)'s status on Tuesday, 25-Feb-2025 17:16:35 JST Tyler
I love the postgresql documentation btw In conversation permalink † top dog :pedomustdie: likes this. -
† top dog :pedomustdie: (dcc@annihilation.social)'s status on Tuesday, 25-Feb-2025 17:17:28 JST † top dog :pedomustdie:
@tyler It won't In conversation permalink -
† top dog :pedomustdie: (dcc@annihilation.social)'s status on Tuesday, 25-Feb-2025 17:18:12 JST † top dog :pedomustdie:
@tyler O it finished lol In conversation permalink -
† top dog :pedomustdie: (dcc@annihilation.social)'s status on Tuesday, 25-Feb-2025 17:18:29 JST † top dog :pedomustdie:
@tyler 10 gb free'd up, not bad. In conversation permalink -
Phantasm (phnt@fluffytail.org)'s status on Tuesday, 25-Feb-2025 17:20:21 JST Phantasm
@dcc @tyler Postgres and the kernel will figure that out, it's better not to fight them and just leave them do their own thing with a proper config in Postgres (shared_buffers, effective_cache_size and a few other settings). Postgres always caches data that it's currently processing and the kernel puts frequently requested disk data into the page cache. In conversation permalink † top dog :pedomustdie: likes this. -
Tyler (tyler@nicecrew.digital)'s status on Tuesday, 25-Feb-2025 17:21:45 JST Tyler
Yes but he was doing a one time operation that warranted it. In conversation permalink † top dog :pedomustdie: likes this. -
Phantasm (phnt@fluffytail.org)'s status on Tuesday, 25-Feb-2025 17:25:14 JST Phantasm
@tyler @dcc There's now way to force ram only storage for Postgres. The unlogged table flag just disables the write-ahead-log for that specific table meaning that there's no guarantee data committed to the table will correct after a crash thanks to write buffers and such. In conversation permalink † top dog :pedomustdie: likes this. -
Phantasm (phnt@fluffytail.org)'s status on Tuesday, 25-Feb-2025 17:28:43 JST Phantasm
@tyler @dcc There's a pg_prewarm extension that adds tables to the buffer cache, but Postgres can still decide to evict that at any time, if it decides that other data is more important. In conversation permalink † top dog :pedomustdie: likes this.