[01:19:57] my.juniper.net has been down for more than 12h now [07:41:44] good morning [07:42:09] I am creating a new role/cluster for the irc2001 host, that should replace kraz (hopefully) [07:42:45] the current role name is role::mw_rc_irc.pp, that is not very informative [07:43:39] since we'll keep irc2001 and kraz side to side for a bit, what about a role name like "role::mediawiki::irc_events" or somilar? [07:43:42] any preferences? [07:47:53] (I also like role::mediawiki::irc_events_dispatcher) [07:51:55] ok I'll send a code review with some ideas [07:52:24] role::mediawiki::irc_events sounds good to me [07:53:44] moritzm: I thought about the same, but then I created a cluster caleld "irc_events" for irc2001, so I thought it was a bit overloaded [07:53:51] if not I'll amend the cr to ::irc_events [07:54:12] (don't have any strong pref on naming) [08:00:54] <_joe_> if you want a descriptive name [08:01:06] <_joe_> role::mediawiki::wtf is probably the best one [08:01:51] <_joe_> role::mediawiki::please_use_eventstreams too [08:07:25] ahahahhaha [08:17:42] april 1 gerrit commits [15:02:03] Does anyone object to T245242 before I JFDI? [15:02:03] T245242: Allow !log in #wikimedia-sre - https://phabricator.wikimedia.org/T245242 [15:02:18] +1 [15:05:03] _joe_: your UBN! at https://tools.wmflabs.org/bash/search is fixed :) [15:05:36] <_joe_> bd808: FINALLY [15:05:39] \o/ [15:05:44] developers are the worst [15:08:57] lol [15:17:15] !log Toil reduction: !log messages now work from the SRE team's Freenode channel. [15:17:18] Logged the message at https://wikitech.wikimedia.org/wiki/Server_Admin_Log [15:18:04] \o/ \o/ \o/ thanks bd808 [15:22:32] 👍 [15:55:05] !log (log(n)) [15:55:07] Logged the message at https://wikitech.wikimedia.org/wiki/Server_Admin_Log [15:55:16] feature does not produce the expected results, I'm filing a bug :P [15:55:30] <_joe_> bblack: but it goes to twitter! [15:55:43] lol [15:55:51] and mastodon [15:56:01] and outer space! [15:56:55] <_joe_> I added a couple of saturation graphs at the top of https://grafana.wikimedia.org/d/000000550/mediawiki-application-servers [15:58:03] <_joe_> the first one shows the % of available php workers time we're consuming [15:58:18] <_joe_> if it goes over 100%, we're making requests queue up in php-fpm [15:59:16] saturation of worker threads busy? [15:59:37] _joe_: I wanted to ask about this [15:59:47] is this the same metric that drops out when an appserver is completely full? [16:00:10] 👀 https://grafana.wikimedia.org/d/000000550/mediawiki-application-servers?orgId=1&from=1581479233991&to=1581481808821 [16:00:15] <_joe_> bblack: I'm late for the meeting [16:00:18] https://grafana.wikimedia.org/d/000000550/mediawiki-application-servers?orgId=1&fullscreen&panelId=92&from=1581477084637&to=1581486024286 [16:00:24] ^ interesting past datapoint [16:00:24] <_joe_> I need to sign in to google [16:00:48] bblack: ^5 [16:01:04] is that non-emoji for high-five? [16:01:11] I didn't know you spoke non-emoji :) [16:01:37] I'm versatile [16:01:48] there's also not a great high five emoji [16:02:04] yeah, there is not [16:02:28] there is 👊 to offer a fist bump but the actual name of that emoji is fairly offputting for that use case [16:03:02] (in most clients it's : facepunch :) [16:03:05] better way to suggest a fist bump IMO is to combine 🤜🤛 the right- and left-facing fists [16:03:13] but still is imperfect [16:04:36] many people use 🙏 to suggest high-five but that's really U+1F64F PERSON WITH FOLDED HANDS and includes other meanings like "pleading, praying, bowing, or thanking" [16:05:15] <_joe_> can we just not high-five? [16:05:21] don't you see the total failure of emojis right here right now? [16:05:24] <_joe_> we're not a football team [16:05:26] :D [16:05:34] no one needs to high-five who does not want to [16:06:09] volans|off: fortunately the Unicode Consortium is very accepting of adding new emojis, with a pretty clear set of criteria [16:06:47] the failure being that they are not easily conveying a universal meaning even cross languages [16:07:08] as they are conveying many meanings for each one and everyone reads them in a different way [16:07:32] *even more [16:07:37] _joe_: fingerguns it is [16:08:28] <_joe_> I don't dare asking what that is [16:09:10] <_joe_> volans|off: classical example was the ample use of peach emojis for the atlanta techconf [16:10:00] oh no [16:55:23] day 2, my.juniper.net is still down, good thing we won't have any ongoing issues. As it's the platform to (between other things) track support tickets [16:56:20] ah at least now I sometimes get "Due to technical difficulties my.juniper.net has been taken offline. Please use Case Manager at this time. We apologize for the inconvenience" [17:04:28] <_joe_> I hope you didn't reuse the same password elsewhere [17:05:18] haha no [17:05:54] <_joe_> good [17:06:18] <_joe_> although I'd bet on network issues [17:06:32] they probably are using cisco... [17:21:33] the tests in 'wmflib/spec/functions', specifically the "ipresolve_spec.rb" is that running locally on dev laptops or where CI runs... or both [17:48:43] herron: I fixed one of the issues with puppet in deployment-prep, now we're on to a puppet compiler issue [17:48:48] which I don't think I caused, but I might have... [17:49:33] andrewbogott: nice thanks! two steps forward one step back [17:49:52] I posted the latest error in T243226 [17:49:53] T243226: Upgrade puppet in deployment-prep (Puppet agent broken in Beta Cluster) - https://phabricator.wikimedia.org/T243226 [17:50:08] It doesn't need immediate fixing but I'd like to at least know that I didn't break it :) [20:04:53] godog: I need to specify profile::prometheus::statsd_exporter::mappings for the thanos-be VMs to get puppet running. Any preferred value for that? [20:04:57] (I might just do {}) [20:12:13] godog: and thanos-fe01 is quite a bit more broken, leaving that to you [20:34:17] ema: is traffic-cache-atsupload still a useful VM? (Puppet has been broken there for a few days so wondering if it's already done its job and can be deleted) [21:31:12] https://mobile.twitter.com/Foone/status/1223063275996213248 - splendid [21:31:30] oh no [21:33:48] we patched that during allhands https://phabricator.wikimedia.org/T243979 [21:34:01] 2.219 [21:34:12] Jenkins 2.219, LTS 2.204.2 now disables both UDP multicast/broadcast and DNS multicast by default. [21:38:33] mutante: doesn't stop it being lol :D [21:39:03] yea:) [21:47:53] "You can cause an infinite loop of bandwidth on any network with two Jenkins servers." [21:48:17] am reminded of making two IRC bots flood one another off networks