A more general metric that is useful to watch for is pressure stall information for CPU, IO and Memory. https://docs.kernel.org/accounting/psi.html https://doc

A more general metric that is useful to watch for is pressure stall information for CPU, IO and Memory. https://docs.kernel.org/accounting/psi.html https://docs.kernel.org/accounting/psi.html I made a Prometheus exporter for it: https://github.com/arianvp/cgroup-exporter https://github.com/arianvp/cgroup-exporter