Reboots

[adamb]

Hey everyone! Trying to get some ideas/suggestions for reboots of the cluster. We typically reboot all inhouse/client servers once a week as good practice. We would like to do the same with our proxmox cluster. I was thinking of rebooting one node of the cluster on Wednesday and one node on Thursday. Then having a script on each node which would ping the rebooting node, once back up migrate specific VM's back to that host. This way we don't come in the following morning to all 10VM's on one host.

Is there a better way to do this? Appreciate any and all input!

 

[tom]

I do not understand why you want to reboot your servers without any reason. what do I miss?

 

[adamb]

Because its good practice, we have over 3000+ linux box's out in the field and over the years have found they perform much better with regular reboots.

I take it this is something most do not do on a regular basis.

 

[tom]

A proper configured Linux server does not need reboots to perform well here, but feel free to do it - you will get just a bad up-time statistics but this seems not important for you.

 

[komdat]

Well, first of all I also don't get the reason to reboot on a regular base. We reboot the hosts maybe every one or two years for any reason.

But if you want to do it for performance, just reboot one after another. The guests will come up automatically, and in my opinion those are the machines that eventually would need a reboot. Rebooting the host without the guests doesn't make any sense at all. Or am I missing something?

 

[hec]

I think rebooting servers without any reason is only creating downtime. If there is no problem why do you do it?

I reboot only after kernel updates. I also think that you will loose performance. Some services will have to rebuild or create their cache.