Proxmox 7: pcieport 0000:00:01.0: PME: Spurious native interrupt!

A

ajeffco

Renowned Member
Nov 17, 2014
31
11
73
Orlando, Fl USA
Hello,

I recently clean installed Proxmox 7 onto a 3-node cluster previously running Proxmox 6. One of the nodes that has been running Proxmox for a couple quite some time now is suddenly getting the error mentioned above. The hardware on this node is the same that was running Proxmox 6. No changes to the nodes hardware config at all. There were some changes to the other 2 nodes in the cluster and a move from ceph to ZFS replication, but I do not think those would have anything to do with this error.

The adapter in slot 01 is a 10GTek 10Gb PCI-E NIC Card, comparable to an Intel X520-DA1. It's been in operation under Proxmox 6 since about mid February 2020.

Code: 
lspci -vv -t|grep 01.0
           +-01.0-[01]----00.0  Intel Corporation 82599 10 Gigabit Network Connection

In the forums I found another post (PME: Spurious native interrupt!" Kernel Meldungen) that had the same symptoms and a potential fix. While the command suggested (ethtool -C enp1s0 rx-usecs 0) stops the error from occuring, it does so in what I would call an odd manner for me, because the error only manifests AFTER I start a VM. If I run the ethtool command before any VMs start (tried it in cron and a pre-up interface script) and then start a VM then the errors occur and running the ethtool command has no effect. The ethtool command only stops the error if it is run AFTER VMs are started. And if I do not start any VMs the error does not occur.

This is 100% reproducible. It does not appear to be causing any functionality errors but the logs will fill eventually, as syslog alerts are occurring at a very fast rate.

Code: 
Jul 14 17:11:03 pve-noncrit kernel: [ 4798.947197] pcieport 0000:00:01.0: PME: Spurious native interrupt!
Jul 14 17:11:07 pve-noncrit kernel: [ 4802.973833] pcieport 0000:00:01.0: PME: Spurious native interrupt!
Jul 14 17:11:13 pve-noncrit kernel: [ 4809.054319] pcieport 0000:00:01.0: PME: Spurious native interrupt!
Jul 14 17:11:15 pve-noncrit kernel: [ 4811.064939] pcieport 0000:00:01.0: PME: Spurious native interrupt!
Jul 14 17:11:18 pve-noncrit kernel: [ 4814.096008] pcieport 0000:00:01.0: PME: Spurious native interrupt!
Jul 14 17:11:23 pve-noncrit kernel: [ 4819.150887] pcieport 0000:00:01.0: PME: Spurious native interrupt!
Jul 14 17:11:28 pve-noncrit kernel: [ 4824.211610] pcieport 0000:00:01.0: PME: Spurious native interrupt!
Jul 14 17:11:33 pve-noncrit kernel: [ 4829.274663] pcieport 0000:00:01.0: PME: Spurious native interrupt!
Jul 14 17:11:33 pve-noncrit kernel: [ 4829.276385] pcieport 0000:00:01.0: PME: Spurious native interrupt!
Jul 14 17:11:38 pve-noncrit kernel: [ 4834.341583] pcieport 0000:00:01.0: PME: Spurious native interrupt!

Not sure where to proceed with this, just giving some feed back and hoping this helps someone else. If anyone has any suggestions on something to try, I'm open and willing ;).

Thanks,

Al
 
I've the same issue with 10GTek 10Gb PCI-E NIC Card. Do you have a solution for this?
PVE Version 7.2-11

regards
Michael
 
Not sure if this will work for others, but I went into the bios settings for the 10gb card and disabled all references to PXE and SRIOV. This seems to have solved the issue and no further errors.
 
Appreciate this is an old post, but thought I'd add I'm experiencing the same issue with our BL460c G9's. Interestingly we have two different 10GbE adapters in use across the 8 host cluster:

HP FlexFabric 10Gb 2-port 536FLB Adapter - No issue.

HP Ethernet 10Gb 2-port 560FLB Adapter - Constant syslog alerts.

I'm working through the list of potential fixes/workarounds after updating to the latest (2021) firmware had no effect.

Can confirm the issue only presents after a VM is powered on (I thought I had sorted it initially).

pve-manager/8.4.1/2a5fa54a8503f96d

If I sort it I'll update here.
 
You must log in or register to reply here.
Top Bottom