Feature request: Hibernate VMs on PVE host reboot

[iamanewuser]

The #17 solution is not perfect, I have multiple VMs, Some VMs hibernate successfully, but some don't. It seems that the script on #17 and PVE shutdown process both execute in parallel.

So I use custom command to do this manually.
I created a script file reboot.sh

#!/bin/bash
qm suspend 102 --todisk
qm suspend 101 --todisk
qm guest cmd 100 "suspend-disk"
reboot

Everytime I want to reboot, I execute this script in PVE console.

The VM 100 is windows operating system with QEMU Guest Agent installed, using "guest cmd" to hibernate is a better way.

 

[Neobin]

For reference:
https://bugzilla.proxmox.com/show_bug.cgi?id=2388

 

[EleotleCram]

The answer in #17 creates a parallel job and relies on timeouts, this will not work reliably. A proper way of doing this would be to:

1. Create a script e.g. /var/lib/vz/snippets/pve-hibernate-all:

#!/bin/bash

qm list | grep running | awk -F'[^0-9]*' '$0=$2' | while read -r vm_id; do qm suspend $vm_id --todisk 1; done;

2. chmod 755 /var/lib/vz/snippets/pve-hibernate-all
3. Edit /etc/systemd/system/pve-manager.service and add the the following line:

ExecStop=/var/lib/vz/snippets/pve-hibernate-all

Such that in context it reads like:

[Service]
Environment="PVE_LOG_ID=pve-guests"
ExecStartPre=-/usr/share/pve-manager/helpers/pve-startall-delay
ExecStart=/usr/bin/pvesh --nooutput create /nodes/localhost/startall
ExecStop=/var/lib/vz/snippets/pve-hibernate-all
ExecStop=-/usr/bin/vzdump -stop
ExecStop=/usr/bin/pvesh --nooutput create /nodes/localhost/stopall

(The order is important here.)

Then reload systemd cache (systemctl daemon-reload) or simply reboot to pick up the changes.

After this, all running VMs will be hibernated when PVE stops regardless of how you do it. It works nicely with autostart on boot as well.

 

[Roman Shein]

+1 vote in support of the "hibernation by default" feature.

 

[leeyoo]

The answer in #17 creates a parallel job and relies on timeouts, this will not work reliably. A proper way of doing this would be to:

1. Create a script e.g. /var/lib/vz/snippets/pve-hibernate-all:

#!/bin/bash

qm list | grep running | awk -F'[^0-9]*' '$0=$2' | while read -r vm_id; do qm suspend $vm_id --todisk 1; done;

2. chmod 755 /var/lib/vz/snippets/pve-hibernate-all
3. Edit /etc/systemd/system/pve-manager.service and add the the following line:

ExecStop=/var/lib/vz/snippets/pve-hibernate-all

Such that in context it reads like:

[Service]
Environment="PVE_LOG_ID=pve-guests"
ExecStartPre=-/usr/share/pve-manager/helpers/pve-startall-delay
ExecStart=/usr/bin/pvesh --nooutput create /nodes/localhost/startall
ExecStop=/var/lib/vz/snippets/pve-hibernate-all
ExecStop=-/usr/bin/vzdump -stop
ExecStop=/usr/bin/pvesh --nooutput create /nodes/localhost/stopall

(The order is important here.)

Then reload systemd cache (systemctl daemon-reload) or simply reboot to pick up the changes.

After this, all running VMs will be hibernated when PVE stops regardless of how you do it. It works nicely with autostart on boot as well.

nice but not working for my 8.0.4.
by your idea i change the plug point to poweroff and reboot command

change original poweroff and reboot

mv /usr/sbin/reboot /usr/sbin/reboot_now
mv /usr/sbin/poweroff /usr/sbin/poweroff_now

and eidt bit your code with waiting

cat > /var/lib/vz/snippets/pve-hibernate-all <<EOF
qm list | grep running | awk -F '[^0-9]*' '\$0=\$2' | while read -r vm_id; do qm suspend \$vm_id --todisk 1; done;
while qm list | grep running>/dev/null; do sleep 1; done;
EOF

then rebuild poweroff and reboot with permissions

cat > /usr/sbin/poweroff <<EOF
/var/lib/vz/snippets/pve-hibernate-all
/usr/sbin/poweroff_now
EOF
cat > /usr/sbin/reboot <<EOF
/var/lib/vz/snippets/pve-hibernate-all
/usr/sbin/reboot_now
EOF
chmod 777 /usr/sbin/reboot  /usr/sbin/poweroff

 

[mat0482]

i am new to linux and proxmox, but

pvesh create /nodes/localhost/suspendall

seems to work fine

example

/etc/systemd/system/pve-guests.service

[Service]
Environment="PVE_LOG_ID=pve-guests"
ExecStartPre=-/usr/share/pve-manager/helpers/pve-startall-delay
ExecStart=/usr/bin/pvesh --nooutput create /nodes/localhost/startall
ExecStop=-/usr/bin/vzdump -stop
ExecStop=/usr/bin/pvesh --nooutput create /nodes/localhost/suspendall
ExecStop=/usr/bin/pvesh --nooutput create /nodes/localhost/stopall
Type=oneshot
RemainAfterExit=yes
TimeoutSec=infinity