[08:04:34] <_joe_> grafana black belts: i have a metric that has attached a label "php_version=7.2.34-18+0~20210223.60+debian10~1.gbpb21322+wmf5" [08:04:47] <_joe_> and I'd like to only display "7.2.34" in the label [08:05:14] <_joe_> but If I use {{php_label}} I don't know how to apply a regex [08:05:25] <_joe_> is there a way? [08:14:38] I don't know about grafana, though you can do it at the prometheus expression level with this https://prometheus.io/docs/prometheus/latest/querying/functions/#label_replace [08:23:07] <_joe_> yeah I can do it in prometheus or at the exporter level, even [08:23:16] <_joe_> thanks godog :) [08:23:41] <_joe_> (given the exporter is writtten by us, we might want to do that) [08:26:27] for sure, yeah the label is a little unwieldly alright, as a best practice exporters should change data as little as possible, I'd recommend not changing the label or perhaps adding a new one with the mangled value [08:27:06] <_joe_> yeah in this case it's just the output of the php_version() function [08:27:41] <_joe_> we can simplify it in the exporter in this case - no additional info for us, quite the contrary, we risk having split metrics whenever we rebuild a package [08:33:57] that's fair yeah [08:47:42] yeah, +1. label_replace() is pretty useful some cases, but in this one the original data appears to carry useless info. [10:00:00] hi folks [10:00:13] does anybody know the login details for https://pypi.org/user/mediawiki-utilities/ ? [10:00:25] I'd need to push in a few days a new version of the mwapi python package [10:00:46] and the above user is listed as maintainer (I also see other packages related to it) [10:01:25] (I am also going to ask Aaron more info, and in case I'll save the details in pwstore) [10:02:27] sorry, no idea [10:04:23] last commit from halfak in 2015 [10:04:48] yeah I know [10:04:57] he's the other maintainer [10:04:59] yuvi for the .deb branch [10:05:15] but there are commits from 2022 for the mwoauth package afaics, so probably it has been used recently [10:05:19] I asked to Aaron via email details [10:05:47] (we are adding aiohttp support to mwapi, it is needed for revscoring etc..) [10:06:17] (adding a new Session class, so no current user will be broken with a pip upgrade after the release) [12:56:14] anyone know if there's a way to reset DRAC passwords without physical access? re: https://phabricator.wikimedia.org/T313369 . I thought there was a cookbook but it looks like the cookbooks assume default PW or require you to know the existing pw [13:03:39] hmm, looks like there is a way https://phabricator.wikimedia.org/T313369 [13:03:48] oops, I mean https://www.dell.com/community/Systems-Management-General/Reset-lost-iDRAC9-password/td-p/7821073 [13:07:24] inflatador: yes we have various ways [13:07:51] * volans reading task [13:08:09] ACK, thanks [13:08:29] have you already tried to fix it? [13:08:29] https://wikitech.wikimedia.org/wiki/Management_Interfaces [13:08:45] the remote IPMI issues [13:09:56] why do you think the problem is the password? [13:10:06] inflatador: ^^ [13:10:25] I can get to the SSH prompt, it doesn't accept the pw [13:10:44] have you tried the old one in pwstore? [13:10:47] doesn't work with Dell defaults either [13:10:47] same file [13:11:37] just tried, didn't work (there are pretty new servers) [13:17:46] ack, let me try something [13:24:45] inflatador: I've fixed elastic2067, I'll update on task and related docs [13:25:10] volans that was fast! Thanks [13:57:48] volans: https://wikitech.wikimedia.org/wiki/Management_Interfaces#Local_IPMI links to itself (rather than the question at the top?) [14:01:02] RhinosF1: yeah was fixing them, thx [14:01:12] should be all fixed now, I've also make the commands safer [14:03:13] np [14:47:36] <_joe_> XioNoX, topranks do you see any reason why we're seeing connect timeouts from appservers to poolcounter at random times? https://phabricator.wikimedia.org/T310835 [14:47:44] <_joe_> started to happen in the last few days [14:49:27] _joe_: any more specific source/destination/start time? [14:49:48] if in the last few days it could be the new linecard, but can't be c5 [14:51:09] from the task is it more a June 16th kind of issue or July 14th? [14:51:20] <_joe_> XioNoX: mw1412.eqiad.wmnet to poolcounter100{4,5} [14:51:22] if july 14th it can't be the new linecard [14:52:01] <_joe_> the last flurry of problems was in the last few days apparently [14:52:15] <_joe_> but yeah if nothing springs to mind, I'd leave it as is [14:53:46] mw1412 is in row C same for poolcounter1005, but that seems more of a coincidence [15:51:58] XioNoX: oh, thanks for the correction on that doc -- for some reason I thought some stuff outside of row C was affected too, but I must have been wrong [15:52:08] sorry for scapegoating your perfectly good router! [20:09:51] Not sure if it's interesting to anyone but saw https://blog.cloudflare.com/using-apache-kafka-to-process-1-trillion-messages/ today