[08:17:10] <elukey>	 Hey folks, I noticed an alert related to poolcounter "full queues" (IIUC locks refused because too many workers were working on queues)
[08:17:39] <elukey>	 I got worried a little since we replaced recently the poolcounter VMs to run on Bookworm, but then I checked https://grafana.wikimedia.org/d/aIcYxuxZk/poolcounter?orgId=1&from=now-30d&to=now&var-dc=eqiad%20prometheus%2Fops
[08:18:16] <elukey>	 I don't see a regression, but now I am wondering if the poolcounter node need a quick perf check
[08:18:35] <elukey>	 (not enough capacity etc.., very ignorant about poolcounterd so I can't tell for sure now)
[08:26:01] <elukey>	 even if, after reading https://wikitech.wikimedia.org/wiki/Michael_Jackson_effect (solution), the queue full may also be an issue of too many clients hitting the same page at the same time
[08:26:41] <elukey>	 if anybody has more context lemme know, otherwise I'll open a task for https://phabricator.wikimedia.org/project/view/437/