[09:36:54] hello on-callers [09:37:28] I'd need to roll restart 4 pybals (lvs2010, lvs2009, lvs1020, lvs1019) to pick up a new VIP [09:37:35] will start in a few [09:41:41] of course I made a mistake [09:47:15] I am fixing it with https://gerrit.wikimedia.org/r/c/operations/puppet/+/920649, puppet disabled on 3 lvses affected, only applied to 2010 (codfw low traffic standby) [10:04:56] {{done}} [10:12:36] <_joe_> godog, jayme the evil plan worked [10:12:48] <_joe_> the jobs are correctly relabeled and the dashboards all work [10:13:16] _joe_: sweet! [10:13:52] <_joe_> interestingly, there is a small reduction in the requests/s to the service because we send less useless probes https://grafana-rw.wikimedia.org/d/RKogW1m7z/shellbox?forceLogin&orgId=1&var-dc=codfw%20prometheus%2Fk8s&var-namespace=shellbox-syntaxhighlight&var-release=main&var-service=shellbox&from=now-15m&to=now&viewPanel=86 [10:35:40] https://www.memorysafety.org/blog/sudo-and-su/ attempt to rewrite sudo/su in rust [10:40:08] The security problems with sudo are not because it's C but because the sudoers syntax makes it very easy to accidentally give too much access [10:40:36] _joe_: nice! [10:42:22] klausman: I'd pick a bit from both columns, to be honest :) [10:42:43] * Emperor uses userv elsewhere [10:42:44] I mean, using C surely doesn't _help_ [10:54:30] skimming the 2022/2023 sudo CVEs it seems about half of them were moot with a rust-based implementation (double frees, OOB memory access), with the other half being logic errors or missing sanitising which would also apply to a rust code base [12:55:00] interesting [14:20:58] hello folks, I am investigating an authdns-update issue with su*khe, please hold off to run dns updates for the moment [14:52:55] fixed, you can now proceed :)