CMK version:
Checkmk Raw Edition 2022.11.15
OS version:
Docker on Debian Bullseye
Whenever docker restarts, the growth trend computation of the filesystem check on many hosts does a huge jump, and only slowly (over days) receedes back to its previous, steady value.
I’ve seen these jumps do a huge spike up, a huge spike down, and go to zero.
These jumps have happened after doing “service docker restart” on the host, but also after doing “docker-compose down” followed by “docker-compose up -d” for the checkmk docker instance.
After 2 of each (the second docker-compose down/up just to verify before creating this topic), I can confidently say that this is a reproducible problem.
Besides making these graphs pretty much unusable, it has forced me to disable all other checks on filesystem size beside current usage. Both growth percentage and time until filesystem is full go haywire, producing plenty of false positives, which persist for days.
This bug appears both in CRE 2022.11.15 and in CRE 2022.07.27
Has anybody else run across this? If this is a known bug, any chance of a fix? Is there a good workaround, besides just not checking filesystem growth?
Screenshots of growth graphs:
And here one showing 3 instances of it, just to show this was not an isolated incident:
