Flapping icinga2 clients on windows

Service Monitoring
1

Hi,
we increased the number of monitored services and hosts in the last time from day to day.
A lot of them are windows icinga2 agent based checks.
At some point (exceeding ~350 services / ~65 hosts) we have the issue that from time to time our windows icinga2 clients are flapping to UNKNOWN as they produce
Remote Icinga instance '<client-hostname>' is not connected to '<servername>'
When occurring this happens to a lot of services / hosts at the same time and disappears again for all of them at one time, returning to the usual amount of problem states.

I couldn’t find anything that could explain that. Performance of monitoring server is quite unproblematic (Load, Mem, IO).
Anybody any idea?

Our env:

icinga2 - The Icinga 2 network monitoring daemon (version: r2.11.2-1)
System information:
  Platform: Ubuntu
  Platform version: 18.04.3 LTS (Bionic Beaver)
  Kernel: Linux
  Kernel version: 4.15.0-70-generic
  Architecture: x86_64
Build information:
  Compiler: GNU 8.3.0
  Build host: runner-LTrJQZ9N-project-298-concurrent-0

Thanks a lot,
Matthias

2

If this occurs after reloads than it’s by design.

3

There’s a grace period of 5 minutes after reloads where non-existing connections don’t generate unknown results but a noop. It seems that the agents still don’t have a connection after this period of time.