This message was deleted.

You could try assigning more memory to the pod, might decrease the IO if your sys is swapping, faster disks if its possible, or use topology spread to make use of more nodes thus have more IO throughput disclaimer: not really an expert on topic https://kubernetes.io/docs/concepts/scheduling-eviction/topology-spread-constraints/

I am already using several pools

this one is still open https://github.com/kubernetes/kubernetes/issues/92287

Another opened issue for the same