[09:45:13] errand [10:50:33] Lunch [12:00:51] dcausse: I'm implementing that slot filtering and I started to add a slot set to RevCreate, but I don't think I actually need that, do I ? that slots don't really matter the initial filtering of events, or am I missing something? [12:40:06] lunch [13:01:02] zpapierski: I think the rev slot info only matter for the initial filtering and does not need to be carried around at every step of the pipeline [13:02:38] at a glance only RevisionCreateEvent needs to have this "revslots" map [13:48:39] zpapierski: is https://phabricator.wikimedia.org/T296470 the rollout plan for WCQS? [13:52:04] mpham: yes, that looks like a rollout plan [13:52:38] Since this is a new service, we'll want a few days of just watching it to make sure it works as expected before announcing it as generally available. [13:52:58] makes sense. i'm just updating project tracking (asana) [13:53:19] We do have a few peripheral actions that we might want to list there (or somewhere else): redirect the beta service to the production service for example [13:55:56] as in we might want to add them to asana? [13:56:52] we have them in asana already, so that's probably enough [14:10:18] mpham: what gehel said - this is a rollout plan that needs to happen before being presented ( by which I mean mostly enabling the domain and redirecting the beta endpoint) [14:10:52] I have both of those in the tickets, I didn't feel the need to put them into a separate plan (though I can, if that's helpful) [14:12:56] It is unlikely that we'll forget those steps and they are trivial enough, so they don't affect planning much. [14:13:09] whatever tracking we have on those is most probably sufficient [14:13:09] (that was my thinking as well) [14:59:37] yay, java modules errors, in Scala code [14:59:38] * zpapierski Unable to make field private static final long java.util.Properties.serialVersionUID accessible: module java.base does not "opens java.util" to unnamed module @5c5eefef [15:13:36] gehel: I forgot where to take from the intellij settings for checkstyle, I know you provided me them last time, I'd leave to have them again [15:13:41] s/leave/love [15:59:46] ryankemper: can you have a look at https://phabricator.wikimedia.org/T294154#7534209 ? [16:00:42] I'm surprised to see "buster" in there [16:01:04] zpapierski: https://repo1.maven.org/maven2/org/wikimedia/discovery/discovery-maven-tool-configs/1.17/ [16:52:05] dcausse: I need to coffeinate myself, might be 2min late [16:52:12] zpapierski: np [16:52:44] gehel: good catch. i suspect that was my oversight on the original ticket (I seem to remember the hq request template defaults to buster) [16:52:46] zpapierski: as mentioned during the triage meeting, T294799 might be a good task for ejoseph, in particular while you are on vacation. Maybe dcausse can help provide some pointers into the code and the testing process [16:52:47] T294799: Special:Search requires ugly horizontal scroll on mobile - https://phabricator.wikimedia.org/T294799 [16:53:41] ryankemper: if one of the servers has been reimaged to buster and it just works, let's not complain! The end goal is still to move to bulllseye... [16:53:48] dinner! [17:31:42] dinner [17:33:39] gehel: We'll look at that before I go [17:51:30] ryankemper: wanted to check in and see when we can move wcqs back to lvs prod. Hopefully easier this time at least :) [17:52:37] ebernhardson: ah yes I'll get moving on that today [17:52:50] shouldn't need any pybal restarts or anything so should we pretty straightforawrd [17:52:54] straightforward* [17:53:22] oh good, i was expecting pybal restarts still [18:00:50] Actually I'm probably misremembering, I think joe or someone did pybal restarts the next day so I just wasn't remembering that part [18:01:06] Anyway regardless I know how to roll restarts, just means I'll need to get traffic's approval first [18:48:05] * ebernhardson is not finding any obvious reason prometheus is not collecting jmx data :S [18:48:22] unrelatedly, i wonder how blazegraph on wdqs1012 has 10k open tcp6 connections in TIME_WAIT [20:01:44] cirrus traffic is back on eqiad [20:09:29] \o/ [20:09:33] thanks for handling the backport [20:54:43] meh, should have been obvious sooner. The problem with jmx metrics collection is i pointed the collector at profile::query_service::blazegraph which is a define and not a class, and the puppetdb query here cares about the difference [20:57:22] not entirely sure on right answer..i previously added `resource_config` which issues the correct puppetdb query, but jmx has extra stuff in the middle that makes so we can't swap it in [21:09:21] meh, i suppose the easy answer is to separately declare collectors for wdqs and wcqs. It's not like we plan to have 10 of them [21:53:45] put up a fix for jmx stuff, i think it might work but it makes me suspicious of pcc output. The prometheus config files pcc outputs seem to be missing a bunch of hosts (not just for these, but elastic as well).