[08:42:08] About to do s1 codfw switchover [08:42:21] good luck! [08:46:51] cezmunsta: ok to merge your dbctl change? [08:46:55] db2212 with 0 weight? [08:47:12] ah you merged it already with my change [08:47:48] sry, yes hit enter a little too quick [08:47:54] no problem [08:47:56] it was safe to merge :) [09:00:16] All done [09:58:48] cezmunsta: I added some details to https://phabricator.wikimedia.org/T429455 [10:13:11] federico3: why are those steps on a review ticket? [10:16:15] if you prefer we can move them elsewhere [10:19:47] federico3: Yeah, I think longer discussions on what's needed etc. Let's wait for the review and then we can sit down and talk about its future [10:20:23] we already discussed some of the improvements in meetings [10:21:35] federico3: Yes, but it is better to wait for this review/audit to happen so the three of us can have an idea of its status and where we want for its future [10:22:00] We have all those on docs [10:22:07] ok [10:33:44] I have moved the list to a comment [10:39:26] federico3: have you been able to CI-build wmf-admin for trixie? [10:42:04] moritzm: I think so: https://gitlab.wikimedia.org/repos/sre/wmfdb/-/jobs/862096 ran on trixie and generated https://gitlab.wikimedia.org/repos/sre/wmfdb/-/jobs/862096/artifacts/browse/WMF_BUILD_DIR/ [10:42:55] ok, if these are working fine, then please import it into "main" for apt.wikimedia.org / trixie-wikimedia [11:20:05] FIRING: MySQLReplicaNotUsingGTID: MySQL replica db1182:9104 not using GTID - https://wikitech.wikimedia.org/wiki/MariaDB/troubleshooting - https://grafana.wikimedia.org/d/0fec1d02-1b0b-44c0-84b0-64894f3ba682/mariadb-gtid - https://alerts.wikimedia.org/?q=alertname%3DMySQLReplicaNotUsingGTID [11:20:13] ^ me testing [11:23:37] moritzm: when you have a minute I'm not sure how to make reprepro understand which CI run to pick the packages from [11:28:18] see https://wikitech.wikimedia.org/wiki/APT_Staging_repository [11:40:58] uhm this looks sus https://phabricator.wikimedia.org/P94244 [11:44:50] check with the sre collab folks, they run this setup [11:45:56] thanks [12:10:21] What is this server? labtestservices2001.codfw.wmnet [12:12:26] cezmunsta: We do not own that server, but I think it was decommissioned long ago [12:12:52] https://gerrit.wikimedia.org/r/c/operations/puppet/+/505812 yep, decommissioned 7 years ago [12:13:03] https://phabricator.wikimedia.org/T218022 [12:19:40] Hmm it comes out of SELECT name, port, section FROM zarcillo.instances INNER JOIN zarcillo.section_instances ON instances.name = section_instances.instance ORDER BY instances.name DESC [12:19:58] we can defeinitely delete it from zarcillo [12:20:01] want me to clean it up? [12:20:32] there could be more, one sec... [12:21:32] labtestservices2001, labservices1002, labservices1001 with their sections labtestservices and labservices [12:21:56] we don't own any of those [12:22:01] so we can probably delete them [12:22:40] es1045, db1262, db1224 can't be reached, I think that is all expected, but then... what about the grant that needs to be applied? [12:23:02] yeah those are being worked on [12:23:04] es1045 reimage [12:23:13] db1262 down https://phabricator.wikimedia.org/T428832 [12:23:23] db1224 down https://phabricator.wikimedia.org/T427535 [12:23:55] es1045 is now up [12:29:15] ack [12:55:05] RESOLVED: MySQLReplicaNotUsingGTID: MySQL replica db1182:9104 not using GTID - https://wikitech.wikimedia.org/wiki/MariaDB/troubleshooting - https://grafana.wikimedia.org/d/0fec1d02-1b0b-44c0-84b0-64894f3ba682/mariadb-gtid - https://alerts.wikimedia.org/?q=alertname%3DMySQLReplicaNotUsingGTID [15:56:08] FYI, today I deployed a new version of the "maintain-views" script to all clouddb hosts. it is backward-compatible, but ping me if you see any issues. [15:56:23] I will add some docs to wikitech soon documenting the new features [16:27:17] federico3: only pc2024 shows in Zarcillo for pc4, yet orchestrator shows the real state of 2, with pc1024 present. Any ideas why? [16:28:54] select * from instances where name like 'pc1024%'; -> nothing in the instances table on zarcillo [16:29:10] It is in the masters table though [16:29:22] Oh, no that is a different one [16:29:37] but it should also be in the instances table I think [16:29:47] pc1016 shows in orc as pc6 [16:29:58] there's also instances_view [16:30:34] pc6? [16:31:48] select * from masters where section = "pc4" [16:32:02] shows a host from https://zarcillo.wikimedia.org/ui/sections#pc6 [16:42:12] cezmunsta: [16:42:14] https://www.irccloud.com/pastebin/GY2v29z0/ [16:43:01] other than pc4 having only one instance, I'm seeing pcABCD where D == section number [16:45:09] Yes, I can see that. What I am asking is why that does not represent reality [16:46:19] Also, is the masters table used? [17:01:02] I don't know why it went missing. BTW Are we doing any primary switch tomorrow morning? [17:07:16] I was going to do s1, but that has been done so not from me [19:26:18] cezmunsta: deploying pc hosts isnty very automated so probably I forgot some zarcillo steps a few weeks ago when refreshing all that hw. I'll review and fix tomorrow [19:26:24] Thanks for the heads up [19:27:00] cezmunsta: I don't think we use the masters table for anything else apart from zarcillo internal stuff [19:28:15] it is used for monitoring classification- the master is marked as such for prometheus tags [19:28:43] but yeah, nothing substantial [19:30:28] marostegui: jynus: thanks for confirming +1 [19:31:31] cezmunsta: the minimal docs: https://wikitech.wikimedia.org/wiki/MariaDB/Monitoring#Prometheus the details can be seen on puppet