pvestatd hanging?

[jermudgeon]

linux 6.8.8-2-pve
pve-manager 8.2.4

7-node cluster

I'm having nodes go grey-question-mark in the UI; no problems with corosync or underlying storage. Restarting pvestatd brings back up the UI, except for cpu/load/memory stats. And it tends to keep hanging unless I reboot, after which it will go for days or weeks.

What are the underlying logs I could examine to find out why some nodes end up in this state?

 

[Folke]

Hi,
you can read the latest logs with journalctl -eu pvestatd. If you want to limit the output to the incident itself, you can use the since and until options:

journalctl -u pvestatd --since "2024-07-10 08:00" --until "2024-07-10 12:00"

A common cause for this can be slow or hanging storage or network shares, which will cause pvestatd to take so long to report, that the nodes state will be displayed as unknown