Delay of connection to vm console

[SviridoffA]

problem

On the first launch, the connection is established immediately. But after some time delay starting to grow. With a cluster running for 7 days, connection delay can reach approximately 4 seconds, and continues to increase over time. But after restarting Cloudstack agent on the KVM host where the virtual machine we’re connecting to is located, the delay disappears and the connection becomes instant again. The problem returns over time.

versions

ACS - 4.19.2 (cloudstack agent with the same version), issue was also observed in version 4.19.1 Hypervisor - KVM Storage - Linstor Network - Bridges (vlan isolation)

The steps to reproduce the bug

Just let the cluster run for 7-14 days, then try checking the delay. Restarting the agent will temporarily resolve the issue.

What to do about it?

It would be great if this issue could be resolved, as we’d like to avoid restarting the agents via cron.