PVE acting on its own

[damjank]

Hello all,

I have Lenovo P500 Workstation with PCie NVME, 96GB RAM and 24core Xeon. Now after certain update, and I cannot recall which, PVE is acting strange like it has its own mind. I receive certain containers shutdown, I notice something is not working and I go look and it is shutdown. No log in task history, it is just down, RAM utilization is out of this world, and let us not get into RAM cache Linux management discussion. And SWAP, regardless of the RAM usage or cache setting, it stays in mid 1-3GB and at certain moment it goes to 10-14GB. I cannot fault running VM/LXC, performance is top notch, it is just these strange behaviors, also, I am afraid putting this to work in company where we have 7 high powered servers, running VM that need to be up like, 99.99%. Any thoughts?

 

[r.jochum]

RAM utilization is out of this world

Do you have ZFS, lower the ARC cache then: https://pve.proxmox.com/wiki/ZFS_on_Linux

 

[damjank]

Do you have ZFS, lower the ARC cache then: https://pve.proxmox.com/wiki/ZFS_on_Linux

OK I have reseted this to 8GB now, I had it at 32, still there is the question of shutdown LXC at its own

 

[r.jochum]

Do you see out-of-memory notices in the kernel/syslog? (grep -i "out of memory" /var/log/kern.log)

You should see these also on a connected screen if so.

 

[damjank]

no, no out of memory messages, checked actual log and historical, for last month - no such messages

 

[r.jochum]

Try to grep for pve-container or pvedaemon in syslog anything error like?

 

[damjank]

Anything that I could find is "Aug 29 08:50:27 pve systemd[1]: pve-container@101.service: Succeeded.", that was in syslog.1 - nothing else, in any log in logs directory either with pve* or *cont*

 

[r.jochum]

I'm curios whats the uptime? Did the box reboot and start vm's but not container?

 

[damjank]

uptime is: 9 days 22hrs
If I reboot, all VMs and Containers come online without any issue, they also stop correctly except for FreeNAS as I do not have tools installed..