Check_MK Agent causing high Disk IO

The systemd I/O problem was fixed end of May with version 2.1.0p1 or p2-

If your proxmox graph shows such a high I/O, how long is the runtime of a single agent run?
Also here the question how is the agent executed - systemd/xinetd/ssh?

In my very small proxmox setup i have no such problem - nearly no disk I/O at all.
Between 0,5 and 1 MB/s on the proxmox hosts with 4-6 containers/VMs and all are monitored with CMK agent. Data between booth is synced over ZFS.