Is there an existing issue for this?

• There is no existing issue for this feature

What are you currently unable to do

Have a say in what the OOM killer kills

What do you think would need to be added

I had some VMs randomly in a stopped state and it turned out that my host went OOM :(

Unfortunately when this happens random VMs and LXCs are getting killed.

I would like to request the ability to set an OOM score per instance, as some of my instances are more important than others.

incus config set <instance> limits.oom_priority <value>

Something like this ranging from -1000 to 1000, where a low value is never killed and a higher value is more likely to be killed. Similar to OOMScoreAdjust from systemd.