I'm on harvester 1.4.3 and I'm getting OOM crashes and I'm inferring that the host is instigating it because free memory is too low. Likely my next move is to upgrade the cluster. I haven't been able to pin this down with node or kubevirt metrics, although I need to look at it some more, the host looks like it had free memory. Before the crash there is nothing in kubectrl events. We run with zero swap. This is what I'm seeing on the node:

[11838741.555178] Memory cgroup out of memory: Killed process 31356 (qemu-system-x86) total-vm:15279832kB, anon-rss:8536744kB, file-rss:25736kB, shmem-rss:0kB, UID:107 pgtables:23000kB oom_score_adj:948

It might be worth increasing the reserved memory of whatever VM that is, if you didn't already -- see https://docs.harvesterhci.io/v1.6/vm/index/#reserved-memory (We've had cases in the past where IIRC VMs normally run fine but go OOM under heavy I/O load, somehow due to there not being enough reserved memory)

Thanks I haven't adjusted that. I'll give it that a try.