[08:20:53] effie: i think your editor hates newlines at the end of files [08:21:06] i was just tracking back why there was no newline at the bottom of modules/admin/data/data.yaml [08:21:14] 3 commits in a row by you was the answer ;) [08:21:22] yes, I noticed it last week [08:21:32] you can file a complaint with Microsoft [08:22:34] :D [08:23:00] I need to find the setting in vscode that will leave things as they are [08:23:26] `"files.insertFinalNewline": true,` will do the generally-right-thing [08:24:10] but, what is the right thing these days anyway ? [08:24:24] i can't think of a case where you _don't_ want a trailing newline [08:24:32] I was joking :p [08:24:39] oh but please tell [08:24:49] oh i see :) [08:25:22] effie: i've got one. you probably don't want a trailing newline in a filename ;) [08:25:32] :p [09:08:19] FYI, in 10m debmonitor will be switched to Buster, so there might be some temporary issues [09:58:51] kormat: I'd augment your statement: "I don't see a case where you 8don't* want a trailing newline *in a text file."* [09:59:04] I'll keep my ELF and .so's and JPEGs as-is, tyvm [09:59:33] i knew telling you about WMF was a mistake [10:00:15] I'm having a blast(ing you with technical corrections) [10:00:55] <_joe_> klausman: even for text files, not everyone uses a crippled text editor [10:01:30] My editor handles non-nl-at-eof just fine. But it tries to help less privileged editors. [10:01:56] <_joe_> that seems sensible, yes [12:05:36] moritzm: do you have a one-liner for T223430 to check if a given host still needs a reboot? [12:13:52] you can grep for "md_clear" in /proc/cpuinfo, if it's not present it still needs a reboot [12:14:07] moritzm: great, thanks! [12:14:40] we had a handful of old servers with CPUs which Intel never bothered to fix in microcode updates, but they should all be gone after the recent decoms [15:35:45] volans: the `--timeout` param to cumin, does it cover the connect timeout? [15:47:32] as in ssh connect? [15:49:29] yes [15:54:27] kormat: it passes it to clustershell and IIRC it sets the whole execution timeout [15:54:53] so it should include teh connect too, but I see also in their doc that they have a 10s default value for connect_timeout [15:54:56] (Value passed to ssh or pdsh (integer)) [15:55:41] ahh, i see. ok, that's what was biting me [15:55:46] although we do se ConnectTimeout 2 [15:55:49] on the ssh config directly [15:55:54] *set [15:56:25] kormat: you really should convert that to a cookbook, could take advantage of a lot of retry logic in spicerack ;) [15:56:44] yes indeed [15:56:55] like wait_reboot_since [15:57:19] and actually might also be useful as a use-case to moritzm that is working on a more generic reboot cookbook [15:57:26] given the specific needs you have in DB-land [15:58:12] pre/post reboot commands, icinga downtime that covers the whole operation, etc... [16:02:17] <_joe_> volans: you mean for a single server? [16:02:31] <_joe_> can we please fix/reuse reboot-single? [16:03:24] _joe_: that's part of mor.itz OKR, most functionalities should go into spicerack and I pointed him to your recent patches as he was on vacation [16:20:47] <_joe_> heads up everyone: if mobileapps complains, you can probably just revert https://gerrit.wikimedia.org/r/c/operations/puppet/+/625707 and run puppet on the restbase servers [18:01:06] I have scheduled a downtime for 1 day for mediawiki GET in eqiad [18:01:12] latency [18:02:27] I have changed a setting in php and it has degraded performance badly on mwdebug1001