[debating] Should I run nutd directly on PVE or as a VM/container?

[Keeper of the Keys]

Hey everyone,

I have a UPS connected directly to my PVE host by USB, I would like to monitor it with nutd, my gut is that since nutd is a small daemon and also since if the battery gets low it needs to anyhow do drastic things like shutting down the host it should just run directly on the host instead of in a container/VM but I see a lot of the guides going in the container/VM direction.

What are the pros/cons here?

 

[gurubert]

Why are you even thinking about running this in a container?
Just run it as a normal system service.

 

[Keeper of the Keys]

That is exactly my instinct, but because others were doing it that way (there is even a topic about the LXC route on this forum) I was questioning myself.

 

[gurubert]

Containers may have their (specific) use cases. But this is not one these.

 

[sw-omit]

One pro from running it in a container would be the ability to transfer it more easily to a different node in a cluster, for example if that nodes needs to restart/be off for a bit for maintenance, and you want the monitor to remain active during that time (and only run 1 monitor per cluster to shut down the entire cluster for example.
Another one would be to keep your cluser-server "clean" so that updates and the like (both from proxmox and from the software) can't mess up (some) dependancies for the other.

Just my 2 cents, and I think it still should be fine to install it on the server, especially if you at least once a while verify things work after updates and the like.

 

[Keeper of the Keys]

One pro from running it in a container would be the ability to transfer it more easily to a different node in a cluster, for example if that nodes needs to restart/be off for a bit for maintenance, and you want the monitor to remain active during that time (and only run 1 monitor per cluster to shut down the entire cluster for example.
Another one would be to keep your cluser-server "clean" so that updates and the like (both from proxmox and from the software) can't mess up (some) dependancies for the other.

Just my 2 cents, and I think it still should be fine to install it on the server, especially if you at least once a while verify things work after updates and the like.

If it were a networked UPS but here it's just USB so if the host is down there is no monitoring unless I physically plug it in elsewhere, for me the way to move easily would probably just be to build an ansible automation, the server doesn't store anything as far as I recall and if I had clients they would be pointed at a hostname which I could point wherever the server is.

/Edit - thanks to both of you for your input I should have said that earlier :X