VM Shutdown on PVE Host Reboot

[Sarel Pretorius]

Good Day Everyone

We have a standalone PVE host which, from time to time, and as expected, needs updates installed and a reboot if there are kernel updates.

We have set the HA Settings under datacenter -> Options -> HA Settings to shutdown_policy=freeze.

What I understand under this policy is, when we reboot the PVE Host, it will freeze/pause all running VMs, host will reboot and then unfreeze/resume all VMs when done rebooting, however, a reboot command issued on host, executes shutdown commands to all VMs.

This is not the expected behavior that I understand from the shutdown_policy.

If I am misunderstanding this, please can someone explain to me this setting and how to rather pause/freeze VMs when rebooting the PVE host?

Your help will be appreciated

Regards

 

[Sarel Pretorius]

Bump

 

[tom]

I do not understand your setup.

You have a standalone host and you try to configure HA? Not possible.

HA needs at least three nodes cluster, see https://pve.proxmox.com/pve-docs/pve-admin-guide.html#chapter_ha_manager

 

[Sarel Pretorius]

Hi Tom

No, I do not want to configure HA as we only have 1 Proxmox host

What I would like to happen - as with Windows Hyper-V Server - is when you reboot the Host, the VMs suspend/pause and once host has restarted, VMs resume/unpause

 

[dcsapak]

is when you reboot the Host, the VMs suspend/pause and once host has restarted, VMs resume/unpause

that is not (yet) implemented, the ha shutdown policy is for something else (how the services gets treated when powering off/rebooting iow. if they should get moved to another node or not)


what you can do is scripting it yourself, basically:

for i in VMLIST; do qm suspend $i --todisk  1; done
poweroff/reboot

all vms that are set to 'start on boot' should resume when the host comes back up

 

[Sarel Pretorius]

Thanks Dominik

Ill script it as you suggest - thanks for your help

 

[victorhooi]

I have something similar - we have a 4-node Proxmox/Ceph cluster, and are not using HA currently.

I need to reboot nodes for things like kernel updates.

Each of the 4 nodes will have some running VMs, and some stopped VMs.

How do I pause all running VMs, then have those ones automatically resume when the node has finished rebooting?

(None of them have Start to boot set - as people are using them for ad-hoc testing etc - I just need to reboot the node, and have the list of running VMs be the same afterwards).