2023-01-04T00:27:57 *** teepee_ is now known as teepee 2023-01-04T03:42:06 hellcp, forum has died again.... 2023-01-04T04:08:51 ☹️ 2023-01-04T04:09:41 do you need help to kick it back to life? 2023-01-04T04:12:01 I don't think anyone is trying to, at the moment?! 2023-01-04T04:14:24 2023/01/04 03:35:28 [crit] 10139#10139: *32419646 pwrite() "/var/lib/nginx/proxy//2/62/0000124622" failed (28: No space left on device) while reading upstream, client: 64.130.121.90, server: forums.opensuse.org, request: "GET /letter_avatar_proxy/v4/letter/c/b2d939/45.png HTTP/1.1", upstream: "http://unix:/srv/www/vhosts/discourse/tmp/sockets/puma.sock:/le 2023-01-04T04:16:22 so the 9GB disk ran full. but logs are only 1GB 2023-01-04T04:16:46 Uhoh! I see there's no profile for this VM in Icinga yet. The only bleeding about space there is with rsync. 2023-01-04T04:17:05 is about& 2023-01-04T04:27:09 bmwiedemann[m], many thanks :) needs more space... 2023-01-04T04:27:54 Now I'm curious to know how much space the old forums was taking. 2023-01-04T04:27:54 old one had only one 10GB disk 2023-01-04T04:27:55 resized discord01 vda to 20GB. Just needs a short reboot 2023-01-04T04:27:55 and with that, forums are back in business for a while. 2023-01-04T04:28:16 seems to have dies again 2023-01-04T04:28:21 bmwiedemann[m], ^^ 2023-01-04T04:28:37 I'm installing updates and it updated discourse-2.8.13+git2.fbd6c300d-lp154.821.5.noarch 2023-01-04T04:28:51 ahh ok 2023-01-04T04:31:42 and after another reboot it is back again 2023-01-04T04:34:50 Very nice! I'm taking this full disk issue as a sign of prosperity. 2023-01-04T04:35:33 prosperity? looking at a dictionary... 2023-01-04T04:36:29 you mean, we had enough nginx log messages to fill up the space? 2023-01-04T04:38:06 Oh, nginx log? Ouch. 2023-01-04T04:38:44 du -s /var/log/nginx/ 2023-01-04T04:38:44 924140 /var/log/nginx/ 2023-01-04T04:41:34 OK, so those errors - and maybe something else - filled up the disk. 2023-01-04T04:42:08 There was a 40MB crate left in /tmp but nothing else 2023-01-04T04:43:42 I wonder whether that's expected from Discourse, or something bad is going on. 2023-01-04T04:44:31 I mean, Discourse on nginx, of course. 2023-01-04T04:44:50 I think, most of the trouble comes from the nginx/puma interaction today 2023-01-04T04:45:46 by writing error log messages, it filled up the disk further 2023-01-04T04:48:31 Let's see if those two misbehave again. 2023-01-04T04:52:43 currently it fills around 300kB/minute that would be 400MB/d. logrotate should compress things well, so maybe 40MB/d are left and at some point it should delete old logs. 2023-01-04T04:53:50 at this rate, disk should not run full in the next 100 days 2023-01-04T04:57:17 looking at while sleep 1m ; do date +%s ; df / ; done - let's see how that looks when europeans wake up. 2023-01-04T05:02:34 Hmm, yeah. 2023-01-04T07:17:27 a European woke up 2023-01-04T08:36:53 Luciano[m]: well "pretty" to me is standard ports, so 80 for HTTP and 443 for HTTPS 2023-01-04T08:37:29 discord01 nginx is listening on 80 2023-01-04T08:38:06 or is it about a different service? 2023-01-04T08:38:19 jenkins-agent 2023-01-04T10:14:46 Luciano: disk usage is rising by 450 KB/min now. Still not bad. 2023-01-04T14:11:54 *** teepee_ is now known as teepee 2023-01-04T18:42:35 I don't really get this push for fedocal - why are we pushing it instead of a proper calendar? 2023-01-04T19:57:52 well, we need more than one calendar, we do have more than one team that has meetings 2023-01-04T19:58:17 also what does proper calendar mean, and how is fedocal not one? 2023-01-04T21:22:34 *** teepee_ is now known as teepee 2023-01-04T23:46:11 acidsys: Not jenkins-agent, actually. Jenkins agents are meant to be workers, so people won't need to interact with it. 2023-01-04T23:46:29 s/it/them 2023-01-04T23:49:56 * Luciano[m] acknowledges discord01's disk usage rise!