All new Registrations are manually reviewed and approved, so a short delay after registration may occur before your account becomes active.
Proxmox in-place upgrade
as I know some people around here are using proxmox, so I'd like to know if anyone has done an in-place upgrade from proxmox 3.x to 4.x and how that has been going?
( wiki article: https://pve.proxmox.com/wiki/Upgrade_from_3.x_to_4.0 )
I have an aged proxmox installation which I like to upgrade, yet there are VMs running in production on it and I'd like to avoid to migrate them to another server with the need for new IPs and such (no failover in place on those).
A short downtime isn't a problem at all, so I'd rather go for the in-place upgrade via apt if there aren't much reasons or bad experiences with this.
for sure I could install the whole hostnode freshly with jessie and proxmox and restore all VMs from previously taken backups, but that probably may require more time, so only considered as fallback.
any comments or suggestions?
Comments
I did two such in-place upgrades just one week ago, in fact. I was worried too that something would go bananas, but I had zero issues.
Now, I only had KVM virtual machines and no OpenVZ containers. Do you have a mix of KVM and OpenVZ, or one or the other? I'm happy that I only had KVM virtual machines as that meant less hassle for me.
thx for the feedback, much appreciated and glad to hear that you had no issues!
I have a bunch of KVMs on that and only one OVZ container.
The latter luckily is more a development box and a small private teamspeak server, gonna try and convert that to LXC... but if that won't work, I'll simply spawn a new lxc-container, so no priority on that one ;-)
did you follow the procedure from the proxmox wiki or something else?
Yep, I followed the procedure in the exact manner they proposed on the Proxmox VE Wiki. I was almost sure there would be at least some type of issue...but nope, none at all!
>
thanks a lot, that makes me much more relaxed and confident in doing it this way ;-)
will see if I can schedule it directly for tonight and give a short feedback afterwards...
I'm still on 3 due to wanting to stay on OpenVZ but its good to know about the upgrade
I was wondering about the same thing. We have a cluster of 3 nodes running that we've been delaying in upgrading. Nice to see there should be no issues.
I played around a lot with LXC lately on newer proxmox installs and do find it much better than the aged OVZ. in my case I'd like to deploy some LXC instances on that host, hence the need to upgrade... and ovz kernel 2.6.32 really isn't something to keep alive at all ;-)
I'll report back here to share more experience with that ^^
I upgraded my proxmox servers, I did choose to do a fresh upgrade though. I will say I had several openvz containers. I did the conversion to LXC and didn't run into any major problems.
Some of the networking configs got a little strange but entering into the container and redoing the network was all it needed.
What aspects are LXC much better than OpenVZ? just curious. I feel strange when LXC is mixed with KVM because it 'occupy' all of the processor
Wish all fine for your proxmox update. I just begin proxmox user from proxmox 4 so I don't have any upgrade experience
I also had no issues as well upgrading Proxmox from 3 to 4 on an old internal server, they were only running KVM however.
yes, while prepping I tried just to restore a dump of my OVZ onto another proxmox 4 server under LXC. exactly as you describe, not much of a hassle, just needed to reconfigure network inside the container afterwards.
first of all it enables running on a much newer kernel on which a lot of features may depend. it seems weird to run something from jessie-backports on a 2.6.32 kernel ;-)
only thing I can say is that I personally feel I can handle LXC well and it fits my use case, those containers feel more lightweight than OVZ ones. but I wont discuss on security and such things as there are a lot of different opinions to this...
that has been discussed here https://forum.proxmox.com/threads/q-lxc-configuration-wizard-what-is-cpu-limit-cpu-unit.22661/
you simply see all cores on the LXC guest because /proc more or less is directly linked into the container. so if doing cat /proc/cpuinfo you don't really see the ressources of the container but the host. but that doesn't mean the guest is able to use all of those...
probably a good point for overselling to unaware customers, haha, tell everybody they get 8 or 16 vCores, which they even can see. yet they may be limited without knowing...
as promised a short conclusion: everything went very smooth, took about half an hour to do all those steps including some last snapshots and the two reboots. no need to restore/redeploy the KVMs at all, it's just working after the upgrade without the need to interfere.
converting ovz containers to lxc also is no problem at all, one only needs to reconfigure network afterwards (virtual mac required for single failover IPs!).
The PMX upgrade is a very smooth process. No issues so far.
I did it, it was smooth. Upgraded debian from wheezy to jessie and proxmox from 3,4 to 4.1 and configured networking to a couple of openvz to lxc. I did it 2 months ago, no issue at all. Just followed the official instructions:
https://pve.proxmox.com/wiki/Upgrade_from_3.x_to_4.0