Page MenuHomePhabricator

replace cloudcontrol1004?
Closed, DeclinedPublic

Description

We're seeing some extremely poor performance on cloudcontrol1004. No guarantee that it's hardware yet but it sure feels like hardware.

This is a pretty old box, and it would be interesting to replace it with an identical spare equally-old box.

Event Timeline

Aaaand, it looks like our problems were the /usr/share/mdadm/checkarray script again like in T224828

Since we disabled the checkarray cron, this began out-performing cloudcontrol1003, which is now showing the same problem because its cron started up before we disabled them. I'll stop the check.

Mentioned in SAL (#wikimedia-cloud) [2020-08-06T20:06:34Z] <bstorm> manually stopped the RAID check on cloudcontrol1003 T259760