You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
PSI metrics are provided on pod level and node level.
For pod level, the value is record in cgroup files under kubepods dir. This is supported for both cgroup v1/v2 for anolis kernel but only cgroup v2 for upstream kernel version.
koordinator supports both cgroup v1/v2 in latest version, i think this can work well in any other OS.
For node level, /proc/pressure/cpu or cpu.pressure in cgroup root dir should be same. Now koordinator only supports cpu.pressure in cgroup root. I'm not sure whether PSI is supported in cgroup v1 for Amazon Linux, maybe you can try cgroup v2 in Amazon Linux.
What happened:
For different OS, the PSI and CPI would have different exposing directory or names. For examples, in Amazon Linux, when querying PSI, the directory is
But on Alios
My question is
1.what's the behaviour for koordinator to pick up different OS?
2.Any additional configuration should do?
What you expected to happen:
Environment:
Anything else we need to know:
The text was updated successfully, but these errors were encountered: