Issues on R86S

[rbe]

Hi,

long term Proxmox User here, now extended to use PVE (currently 7.3-6) on an R86S as OS below my Opnsense router at home. I'm facing a weird issue and got no idea where it comes from..
My opnsense is running just fine (well, it is affected by the current N6005 issues which I believe are fixed by the newer kernel that comes sunday), however as soon as I start a 2nd VM, no matter if it has a network card or not, the whole node seems to freeze for about a minute, then returns for a few minutes and freezes again.

The only thing I can find in the log during that time is:

Feb 13 19:29:31 host pve-firewall[2145]: firewall update time (86.640 seconds)
Feb 13 19:30:02 host pve-ha-crm[2258]: loop take too long (90 seconds)
Feb 13 19:30:02 host pve-ha-lrm[2274]: loop take too long (101 seconds)
Feb 13 19:30:03 host pve-firewall[2145]: firewall update time (22.636 seconds)
Feb 13 19:30:04 host pvestatd[2147]: status update time (128.855 seconds)

My config is rather simple, base OS install + one vmbr that is connected to one of the 10G SFP+ interfaces.
The graphs on the UI are also showing a gap in the timeframe where I face the issues after the 2nd VM runs.
As soon as I shut down the 2nd VM, everything returns back to normal and seems to run fine again.

Any idea where I could start searching for the issue root cause?

 

[Richard]

Hi,

long term Proxmox User here, now extended to use PVE (currently 7.3-6) on an R86S as OS below my Opnsense router at home. I'm facing a weird issue and got no idea where it comes from..
My opnsense is running just fine (well, it is affected by the current N6005 issues which I believe are fixed by the newer kernel that comes sunday), however as soon as I start a 2nd VM, no matter if it has a network card or not, the whole node seems to freeze for about a minute, then returns for a few minutes and freezes again.

The only thing I can find in the log during that time is:

My config is rather simple, base OS install + one vmbr that is connected to one of the 10G SFP+ interfaces.
The graphs on the UI are also showing a gap in the timeframe where I face the issues after the 2nd VM runs.
As soon as I shut down the 2nd VM, everything returns back to normal and seems to run fine again.

Any idea where I could start searching for the issue root cause?

"Freezing" (precisely rather "very slow") may be caused by huge amount of I/O delay. Statistic figures of it are shown when selecting in WEB GUI "Summary" of a node. High I/O delay can be caused by access problems to a media or simply intense write/read operations to it.

 

[rbe]

Yeah, I thought of that too at first but the Opnsense VM is hardly doing anything and the whole system is running on an NVMe SSD using ZFS so basically idle.
Is that maybe one of the side effects from the bugs that currently exist for the Intel N6005 CPUs?

 

[Richard]

Yeah, I thought of that too at first but the Opnsense VM is hardly doing anything and the whole system is running on an NVMe SSD using ZFS so basically idle.

I/O delay is not necessarily caused by "doing anything", but just slow (or defect) access to periphery.

Is that maybe one of the side effects from the bugs that currently exist for the Intel N6005 CPUs?

I don't know since I don't know about these bugs either - do you have more information about them?

 

[rbe]

I was referring to this bug: https://forum.proxmox.com/threads/vm-freezes-irregularly.111494/
However it looks like mine is actually solved after installing the intel-microcode package. I'll leave the 2nd VM running for a few days and reply back here again.

Thanks for responding on it Richard!