It was a cp --reflink=auto operation on the PVE 9.0 host. I wanted to move some files between ZFS subvols with cp --reflink=auto and then delete the source (I think it's faster than mv or rsync). However, operation took all the host CPU usage and then all LXC / VM stopped working due to the lack of CPU cycle. The zfs pool is 2x HDD mirror plus 2x SSD special device mirror, and the host is proxmox-ve: 9.0.0 (running kernel: 6.14.11-4-pve).
Any idea why the cp operation took all the CPU cycles? TIA.
Any idea why the cp operation took all the CPU cycles? TIA.