Is there any new information on this? We have several systems with many small VMs that we monitor with cmk.
The problem with this is that this kills basically smaller installation with low-budget server systems which now require high end DC NVMEs to handle the IO load since all systems running the cmk agents are constantly causing them. Good luck explaining this to a customer, because it worked all those years without issues. And suddenly they are supposed to spend a lot more on the storage devices for monitoring applications?!
Additionally this is causing unnecessary IO on network storage systems putting it under higher load for no real reason.
When stopping the check_mk* services the IO drops to almost zero. Attached is a screenshot from the hypervisor (proxmox) vm disk IO stats.
How do we go from here? 