Restarting a customer's VPS: hard reset vs. clean reboot?
Just had a hoster do a hard reset on my VPS, because they apparently needed to reboot it because of some network config changes.
Well, maybe, but why do a hard reset (unclean shutdown) instead of doing a clean shutdown/reboot?
Have you experienced similar? Do you care?
Thanked by (1)Not_Oles
Comments
Quite common (not the actual reboots per se but that this "unclean" shutdown happens). Many times it's also because I don't always run the qemu guest agent within my VM (but it is ACPI enabled, so that should work). I of course don't like it at all but usually nothing bad comes of it (other than a fsck at startup).
I've made my peace with it (since it doesn't happen too often).
Did you have qemu guest agent installed? (or agent for their platform)
Maybe clean reboot was not possible.
No.
ACPI is sufficient for that, and clicking the "Stop" button on the panel does actually do a clean shutdown.
So this happens quite a bit. Less organized hosts do it more often because even their services crash and need a reset sometimes.
This is why it’s important to use containers. I like easypanel because their free tier lets me do so much for free and manages it all, and I don’t apply most of my tech skills. If server faults, it first relaunches easypanel and then easypanel begins to auto-restore services and you can watch with logs. This is really nice IMHO.
If there is another control panel that delivers this deal of automation with any free tier, let me know as I would love to find a real competitor to this.
If I have an unclean shutdown it's 99% of the time going to be my UPS running out and not being able to shut the systems down in time. The other 1% is a kvm guest getting hard-locked and not responding to normal shutdown commands. This will hang the running agent and you'll end up having to login and do a manual qm stop. If that doesn't work then you start worrying about filesystem corruption doing a kill -15 / kill -9 on the running process. If that doesn't work, scheduled reboot between 1am and 2am the next day.
I run about 15 proxmox nodes and I've only had that process issue happen once in about 2 years.