00:12:41 o0x1eef: that's pretty amazing. i'll have to ask sometime how hard they were to setup 00:24:56 It was a bit challenging but overall I think it was worth it. Most services run in jails and that's relatively straight forward. The most complex piece was ollama because I run it in a Linux VM since the nvidia driver support is more complete with CUDA support. 01:00:30 polarian: you must have a shitty isp because my selfhosted things are rock solid 01:11:58 rtprio: no I have a good ISP 01:12:00 but most dojt 01:12:51 radhitya: self hosting email tends to be forbidden, and server use often comes under commercial terms 01:13:23 I used to self host on a domestic virgin media line and reading the tos I was technically breaching their terms... 01:13:56 didnt stop me ofc ;P 01:15:44 ive self hosted for 6 years now (dont get me wrong im still young and not as experienced as most people here) but I can say this... when it comes to mission critical software, such as email or your primary source of communication (in my case XMPP) downtime/server failure means you are fucked... 01:16:05 you could be on holiday and have a disk fail and you wouldnt be there to fix it 01:17:34 a terminal server can help for failed updates done remotely... and if you got friends/family with sysadmin experience you can always call them to be virtual hands (I am thankfully able to do this) 01:18:10 but in the end a hobby turns into a full time commitment, which I have neglected for a year and my infra is falling apart... 01:20:19 dont get me wrong im all for self hosting... but its a commitment, not a choice to be made lightly... and bare in mind its hugely more expensive... my infra costs me ~£60/month for the business broadband + power draw, there is one off payments such as replacement parts or upgrades... in contrast you can get a dedi in the cloud with gigabit symmetric fibre for as cheap as £30-£40/month which is more powerful than even the new server I have just buil 01:21:59 now if you are piggy backing your current ISP and not having a separate line then that drops the costs considerably, but in my experience paying for business internet with a high quality ISP may cost more than average, but its the best choice ive made 01:26:08 o0x1eef: things like media servers for home streaming or backups are easy and effortless to maintain, they go down you arent fucked... 01:26:40 is the mail server at 127.0.0.1 or in the cloud? 01:27:53 apologies for the wall of text 01:30:19 (note: I build servers using old hardware lying around, they are nothing more than repurposed desktops, if you pay for newer components you can rack up a 4 figure price easily) 01:30:43 a RPI is a self hosters best friend still... 01:33:53 so are intel NUCs, companies love them for front desks and stuff as they can be VESA mounted for clean setups for data entry... but the older ones get sold off for cheap... last I checked coffee lake nucs were going on auction for cheap on ebay... slap a hefty ssd into it and you can run your entire infra on it, low power draw and small, only downside to them is they are rather loud at load due to the small heatsink, and they need regular cleaning, 01:33:53 every 6-8 months I have to clean out NUCs used by my family as they get clogged with dust easily and then poweroff from thermal protection 01:49:59 if it's 'mission critical' then the hoster can fuck it up no matter where it's hosted 01:50:11 also what the shit are you self hosting that's mission critical? 01:56:57 polarian: hydroxide basically provides an imap and smtp server over protonmail's HTTP API :) So that one requires the cloud. 01:58:34 it sounds like polarian is saying you need constant maintenance on such things 02:05:06 I don't neccessarily disagree but I think it can be useful to have a home lab / network where you can experiment with ideas before deploying them at scale 07:20:01 Hi, does anyone have pkg.freebsd.org mirrors that are reasonably fast in mainland china? 08:06:04 the i915kms package seems broken on 14.2 08:07:35 urgh, crap, now i can't get into a working system 08:13:07 runxiyu: https://www.freebsd.org/releases/14.2R/errata/#open-issues ? 11:31:22 I'm using daemon(8) to run a go program. the program runs happily, but refuses to stop. 11:31:50 using `-P $pidfile` or `-p $pidfile` makes no difference, just different variations on *not* stopping 11:31:58 has anybody run into this before? 11:32:33 https://git.sr.ht/~dch/ports/commit/93889b4e25856c3991d08f31875302316bca3a1a#www/anubis/files/anubis.in for reference 11:37:23 Did you try to use -p instead ? 11:38:17 Oh sorry. I saw you did. :) 11:40:03 Is it possible there is a trap setup for SIGTERM that aborts the default behavior ? 11:50:58 so its daemon itself that has the problem AFAICT 11:51:40 when an rc.d script tells it to shutdown, with `-p` the child exits, but daemon remains, and promptly restarts the child (grrr) 11:57:28 I've tried `-P`, `-p`, both together, and neither, all unsuccessfully. 11:59:49 ahaha https://reviews.freebsd.org/D47005 it looks like it might be fixed in current but not backported to 14.x yey 12:00:44 hm, with -p the rc script would kill the child and the --restart would lead daemon to restart it, but with -P the daemon supervisor process should recive the shutdown signal, forward it and don't restart the child (well, that isn't exactly specified in the man page, but how else would you terminate the child with --restart?) 12:01:20 nimaje: that is the correct reading via the manpage, yes 12:10:27 anyway I'm 67% sure this is a bug fixed in CURRENT & 14-STABLE but not in RELEASE yet 12:15:36 Yep I think bc1dfc316a2bba97773a14b96f5e976a52524be4 is the fix 12:29:03 that said I dropped `daemon(8)` from 14-STABLE on top, and its still not DWIW 12:31:25 e.g. 12:31:41 https://www.irccloud.com/pastebin/Jub5dIhS/with_pP 12:33:38 https://www.irccloud.com/pastebin/ZFS6o4BK/with_p_only 12:34:24 https://www.irccloud.com/pastebin/ZFGpvb1x/with_P_only 12:35:57 https://www.irccloud.com/pastebin/qSyeMEKX/with_neither 12:36:11 I don't think its fixed in 14-STABLE either then. time for a PR. 12:50:23 and not in CURRENT either. maybe I'm just holding it wrong in rc script. 13:24:29 Maybe try confirm sending SIGTERM to the process running normally quits as expected, then you could isolate the problem to daemon 14:05:27 dch: sorry, just woke up, but re: daemon(8), you *did* test stable/14 and your problem seems resolved there, or no? 14:05:42 It's a lovely spring morning (here). I'm having coffee, outside. Nice. What do I do? run "ansible-playbook gateway.yml" <== all the configs. Let's see how this goes. Lately, I've been running just single updates (e.g. --tags=pf). 15:23:07 i'm trying to change my laptop backlight brightness.. 15:23:32 acpi_video loaded 15:24:19 using i915kms btw 15:24:49 sysctl: oid 'hw.i915kms.enable_dpcd_backlight' is a read only tunable 15:30:32 oh looks like backlight(8) works... odd 16:27:59 runxiyu: see backlight(8) 16:28:10 polarian: yep, saw; thx 16:28:27 i just didn't expect there to be a literal command for it out of the box 16:28:33 welcome to BSD 16:28:40 the base system can do a lot 16:31:35 yay 20:48:16 rofl 20:48:41 just saw this dude named FreeBDSM in some other channel 20:48:46 still laughing