[05:18:36] <_joe_> kormat: for added evil, please check the sha1 checksum of the script when you download it [05:48:45] going to start esams network maintenance in the next 10min [05:53:27] _joe_: sha1, hmm. should i embed the checksum into the script itself? ;) [06:27:16] (with a note saying any changes to this file require 100k cpu years to update the checksum) [06:27:50] haha [09:51:58] kormat: let me know when you think that feedback for https://gerrit.wikimedia.org/r/#/c/operations/puppet/+/601761/ would be useful (people tend to ignore anything with WIP in the title otherwise :-) [09:53:12] ahh, maybe i gave mixed messages. the intention was the communicate that now would be a good time to get feedback, but i know that it's not quite ready to merge yet [09:53:37] (at a minimum, i need to test a bunch of corner cases before i can be confident it DTRT) [09:53:51] ok, that makes sense :-) I'll have a look at it sometime the afternoon [09:54:03] great, thank you for your sacrifice :) [10:07:33] <_joe_> I just had a strange behaviour from puppet-merge [12:02:26] good afternoon, I crafted a patch to update the Doxygen Debian package at https://gerrit.wikimedia.org/r/#/c/operations/debs/doxygen/+/599094/ and could use it to be build and uploaded to buster-wikimedia component/ci :] [13:25:34] is there any way in gerrit to see a raw file in a CR? [13:26:09] kormat: the gitiles link up top [13:26:29] then there's a 'txt' link at the bottom of that [13:27:03] which.. gives you a base64 encoded version. wow. [13:27:29] there's also a 'download' link in gerrit which gives you a .zip I think [13:27:42] oh, and there's `git review -d xxxxxx` ofc [13:28:33] ok, so 3 options which are all awkward. i should have expected :) (thankfully i _can_ c&p from the gittiles file view page) [13:29:47] a +1 with no comments? wow [13:29:53] your reward will be getting many more CRs [13:29:55] ;) [13:30:15] i didn't review the .pp file, as i don't speak puppet yet [13:30:28] but the .sh file was... relatively non-terrible. :) [14:54:49] qq about ganeti in eqiad - I am a little stuck in creating two new VMs to replace archiva1001 and matomo1001 (buster upgrade), as far as I can see we are running a little low on free ram on a couple of nodes [14:55:17] so the instance creation fails.. should I wait/postpone until https://phabricator.wikimedia.org/T228924 is done? [14:55:25] (rack/setup/deploy of the new nodes) [15:05:31] <_joe_> elukey: non-authoritative, but yes, you should [15:07:45] ack thanks [15:18:33] elukey: matomo has no real dependency on eqiad? you can do matomo2001, we have sufficient capacity in codfw [15:23:03] moritzm: I could yes, for some reason it feels weird but you are right, it shouldn't be a problem (Daniel also suggested the same earlier on) [15:28:28] we have that in other places as well, e.g. deneb is in codfw, or irc.wikimedia.org [15:28:54] yes yes makes sense, will change the ip then [15:29:06] for archiva though I'd say that we should stay in eqiad [15:31:34] eventually we should have both matomo and archiva in some form of dual DC setup, but that's not for right now :-) [15:49:57] <_joe_> archiva, yes [15:50:02] <_joe_> matomo, maybe not [15:50:07] <_joe_> why duplicate the horror [15:50:55] <_joe_> also, matomo has a huge db, I would advise against replication there [17:09:35] _joe_ not anymore, it used to be big but now is tiny [17:09:50] finally we have only very small websites using it