Temporary Performance Issues due to network?

[sshaikh]

I usually run a couple of desktops on a Proxmox host running on a Haswell Xeon with 32GB ram. Hardware capabilities aside, the setup has been running well for me. The host is part of a two box cluster (which I use for the datacentre UI and not resilience).

For the last three days or so my desktops became a bit sluggish and unusable. This was across both that I had running. Since it was both, I suspected the host, and even things like ssh'ing into the box felt a little odd. I update at least once a month, so had some updates pending. A reboot after an update didn't help.

All of a sudden, they're now okay and everything is dandy again.

I have been messing with my network this week, including the reassignment of the Proxmox boxes' IP addresses. I also took the opportunity to switch them to DHCP using the instructions here: https://forum.proxmox.com/threads/proxmox-5-3-dhcp-for-vmbr0.53285 (including updating the hosts file which seems to defeat the point of DHCP, but hey).

Is this something that could have caused the issues I was seeing? Perhaps the network needed to "settle" or something? The question is academic now that all is well, but it's worth knowing if such a situation could occur.

 

[Alwin]

Is this something that could have caused the issues I was seeing?

This is a ¯\_(ツ)_/¯. It could have been a multitude of things. Always hard to tell afterwards.

 

[sshaikh]

I figured it was a long shot, but always worth documenting these data points.

If I see the same hostwide performance again, what's the best way to quantify what's going on? I didn't see anything unusual using top on the host for example.

 

[Alwin]

That's a, it depends. You will need to check all endpoints of your workload/connection to see where it might be stuck.