| 25 Nov 2024 |
 @hexa:lossy.network | tremont cores | 17:45:28 |
| @hexa:lossy.network | yeah, never buy appliances 😄 | 17:45:37 |
| @hexa:lossy.network | they tend to sell you shit | 17:45:47 |
| @hexa:lossy.network | alder lake celerons have avx2 again | 17:46:27 |
| @hexa:lossy.network | https://en.wikipedia.org/wiki/List_of_Intel_Celeron_processors#Tremont_based_Celerons | 17:46:57 |
 victorbjelkholm | Eh, it was cheap enough for my purposes so ain't complaining about it :) Had to upgrade the RAM and yeah, no AVX :P Besides Frigate>Tensorflow trying to use AVX (which can be worked around at least), haven't really come across any issues with it which I guess is the main point for me | 17:49:02 |
| @hexa:lossy.network | frigate doesn't pin tensorflow either, so it could require avx at any time | 17:50:59 |
| @hexa:lossy.network | https://github.com/blakeblackshear/frigate/issues/7035 | 17:51:22 |
| @hexa:lossy.network | they didn't care about this problem in 2021 and they sure as hell won't care going forward | 17:51:44 |
| @hexa:lossy.network | so they import tensorflow for the cpu detector | 17:53:21 |
| @hexa:lossy.network | you could be fine with configuring any other detector, maybe | 17:53:27 |
| @hexa:lossy.network | and also audio processing uses tensorflow | 17:53:42 |
|  Jalil David Salamé Messina joined the room. | 18:27:57 |
| Jalil David Salamé Messina | Hi, I'm running into a weird bug/interaction between hass and nginx: when I connect to it locally (over an ssh tunnel) it works, but when I turn on the proxy and connect through nginx, nginx pins a cpu core to 100% and stops working.
I'd like to have a minimal example of it failing, but because my cache and a few other vital services are running in the same service/nginx config, I am kinda hessitant to turn stuff off.
The journalctl and /var/log/nginx logs are unhelpful.
I tried looking around but I cannot find any references to this behaviour. | 20:02:35 |
| Jalil David Salamé Messina | So, I created a coredump of the nginx worker process, and apparently it is stuck during zstd compression. I am 90% this is the cause, but the harmonia cache needs zstd, so I will need to figure out how to turn it off for hass | 21:48:20 |
| Jalil David Salamé Messina | Yup, setting zstd off; in the extraConfig of the vhost fixed it 😅 I'll make a bug report on nixpkgs but I don't know if it is actionable | 21:53:16 |
 symphorien | does the journal contain kernel oops about io_uring ? if so it's a kernel bug | 21:54:42 |
| @hexa:lossy.network | I have the same issue with an nginx instance that proxies harmonia | 21:55:09 |
| @hexa:lossy.network | cc Mic92 | 21:55:12 |
 Mic92 | Well. Good that I just removed this requirement from harmonia yesterday! | 21:56:16 |
| Jalil David Salamé Messina | In reply to @symphorien:xlumurb.eu
does the journal contain kernel oops about io_uring ? if so it's a kernel bug I don't see any io_uring stuff, but I might be looking at the wrong place (journalctl -g io_uring) | 21:57:00 |
| Jalil David Salamé Messina | Nothing in dmesg either (sudo dmesg | rg -i 'oops|io_uring') | 22:00:27 |
| Jalil David Salamé Messina | coredump stack trace: Using host libthread_db library "/nix/store/pacbfvpzqz2mksby36awvbcn051zcji3-glibc-2.40-36/lib/libthread_db.so.1".
Core was generated by `nginx: worker process'.
#0 0x00007f57b9a2bc97 in ZSTD_compressStream2 ()
from /nix/store/6avf0gmd2sl573bv6qhsqybgvf86wdq5-zstd-1.5.6/lib/libzstd.so.1
I'll go to bed now and report the bug tomorrow | 22:02:02 |
| Mic92 | hexa: https://github.com/nix-community/harmonia -> https://github.com/Mic92/dotfiles/blob/0b146b64f02ac88c7a946413ead6b1604a750dd9/machines/eve/modules/harmonia.nix#L5 | 22:10:21 |
| Mic92 | Phaer already tested it today | 22:10:37 |
| Mic92 | but since I rewrote a lot another tester before a release would be nice | 22:10:51 |
| @hexa:lossy.network | I just bumped my infra to the current head | 22:11:25 |
| Mic92 | No more linked C++ dependencies and builtin zstd compression | 22:13:35 |
| 26 Nov 2024 |
 spacekitteh | ipset v7.22: The set with the given name does not exist
otbr-agent[84842]: 00:00:00.033 [I] Platform------: Execute command `ipset flush otbr-ingress-allow-dst-swap` = 256
otbr-agent[84842]: 00:00:00.033 [I] Platform------: Got an error when executing command `ipset flush otbr-ingress-allow-dst-swap`: `Resource temporarily unavailable`
hmm. i have no clue about ipset, nor this message
| 00:11:45 |
| @hexa:lossy.network | and ipset is just a data structure that holds ip addresses | 00:37:09 |
