User avatar
Norm (ノーム) mikoto_bed @norm@shrimp.biribiri.dev
Admin
Private
This account is on an iceshrimp.net instance running on a Raspberry Pi 4.
Location Academy City
main @mikoto@akko.wtf
avatar src https://twitter.com/yoshie263/status/1990971486732165200
banner src https://www.pixiv.net/artworks/125853123
User avatar
Norm (ノーム) mikoto_bed @norm@shrimp.biribiri.dev
6mo
@chjara yup
User avatar
Norm (ノーム) mikoto_bed @norm@shrimp.biribiri.dev
6mo
meow
User avatar
Norm (ノーム) mikoto_bed @norm@shrimp.biribiri.dev
6mo
trying a 32k recordsize now, and basically rewrote the files by doing a VACUUM FULL

the compression ratio has moved up from 1.99x to 2.18x so this might be a better option
User avatar
Norm (ノーム) mikoto_bed @norm@shrimp.biribiri.dev
6mo
this instance just runs on an lxc container in proxmox
User avatar
Norm (ノーム) mikoto_bed @norm@shrimp.biribiri.dev
6mo
moved my postgres data dir to a zfs dataset with recordsize=16k for this instance

for reference, the volblocksize for the akko.wtf vm is also 16k
User avatar
Norm (ノーム) mikoto_bed @norm@shrimp.biribiri.dev
6mo
test
User avatar
Norm (ノーム) mikoto_bed @norm@shrimp.biribiri.dev
6mo
ok, interesting thing I've observed

restarting akkoma makes the outgoing queue go fast, that is until someone makes a post that has to be federated, and then elixir process keep timing out once again
User avatar
Norm (ノーム) mikoto_bed @norm@shrimp.biribiri.dev
6mo
gonna hang out on here for a bit until akko.wtf finally catches up
User avatar
Norm (ノーム) mikoto_bed @norm@shrimp.biribiri.dev
6mo
@izzy I swear the powermac g5 is still one of the best looking computers out there even today
User avatar
Norm (ノーム) mikoto_bed @norm@shrimp.biribiri.dev
6mo
food, meat @kura nice
User avatar
Norm (ノーム) mikoto_bed @norm@shrimp.biribiri.dev
6mo
me tests backup script
wonders why it doesn't work
thinks it's selinux shenanigans once again
turns out the script just didn't have correct permissions set
User avatar
Norm (ノーム) mikoto_bed @norm@shrimp.biribiri.dev
6mo
gonna expand the virtual disk a bit once I'm done the import, assuming it finishes successfuly
User avatar
Norm (ノーム) mikoto_bed @norm@shrimp.biribiri.dev
6mo
seems like index rebuilding takes a TON of disk space
norm@othinus:~$ df -h
Filesystem               Size  Used Avail Use% Mounted on
/dev/mapper/fedora-root   74G   54G   20G  74% /
User avatar
Norm (ノーム) mikoto_bed @norm@shrimp.biribiri.dev
6mo
decided to have the media on the ssd just as before since the ssd is pretty large (2 TB) and also the hard drives are a bit old so I might look at replacing those first when the opportunity arises before storing important stuff on it
User avatar
Norm (ノーム) mikoto_bed @norm@shrimp.biribiri.dev
6mo
gonna try using pg_dumpall to make it easier to do a quick restore of everything without having to recreate the user and whatnot manually, let's see how it works out
User avatar
Norm (ノーム) mikoto_bed @norm@shrimp.biribiri.dev
6mo
@volpeon still can't believe they did that for oblivion
User avatar
Norm (ノーム) mikoto_bed @norm@shrimp.biribiri.dev
7mo
@akko @yuriposting thx
User avatar
Norm (ノーム) mikoto_bed @norm@shrimp.biribiri.dev
7mo
Would love to try using the hard drives for media instead of the SSDs but last time I tried it does cause quite a bit of thrashing.

Though I suppose it might work a bit better now since I have more ram for the ARC cache, plus there is the option for an L2ARC...
User avatar
Norm (ノーム) mikoto_bed @norm@shrimp.biribiri.dev
7mo
@ezio lines up with what I heard about it
User avatar
Norm (ノーム) mikoto_bed @norm@shrimp.biribiri.dev
7mo
need to use this instance more, but at least for now managed to move it to the main server instead of it sitting on the mini pc which hasn't been doing anything else too useful

now how to repurpose that thing...