PROXMOX cluster node was upgraded and now it do not work in the cluster, anyone knows?

[philjans]

Hi,
We had a crash of cluster (https://www.experts-exchange.com/questions/29198800/PROXMOX-web-interface-don't-work.html) and during the crash, a user did an update on it.
Here's the config of the 4 "normal" node:



I have not the 5th one boot so I cannot run the command on it but it should have the latest version.
I don't know if I boot the 5th is the cluster will go down, anyone know?

What I would expect is that the 5th would simply not be able to "join" the cluster at which point I would be looking to downgrade it to the same versions of the others. Is there a "go back to previous version" with that Proxmox software ?
Thanks for the info

 

[wolfgang]

Hi,

I don't know if I boot the 5th is the cluster will go down, anyone know?

I guess there will no problem.
I the worst case the services can't start, but they will not send evil packages ;-)

But you must finish the upgrade on this node.
How this works depends on where the upgrade stopped.

 

[philjans]

Hi,
It is working: the cluster is up and running with all the nodes
The upgrade was done completely on the 6th node. So I guess, now, I should update the other nodes also?
I'll search for the procedure but if you happens to have the link, it is appreciated.
And can you confirm that when I see "SMP PVE 5.0.18-2" the version is 5.0.18?
Like my last node (the one that was upgraded) shows 5.4.65
thanks

 

[wolfgang]

PVEbld15 is up to date.

then update one by one.
Stop or migrate all VM from the node that you like to update.
If it is clean run the upgrade process.
Then reboot the node.
Then move on to the next node.

 

[philjans]

thanks.
If there is an issue with one node (don't work after the upgrade):
- How can I use another node to go get the vms of the defective node back?

 

[wolfgang]

This dependence mainly on what storage you use.
If you have shared storage or distributed you can move the VM config in proxmox cluster fs.
If not then you must restore a backup, because you need the VM image to run the VM.

Your nodes are on the same release so an upgrade is no big thing.

 

[philjans]

I have a NAS ( FreeNas ).
I know I could move the vm's before the upgrade because I see them, but my question is more if the node pvebld15 is not working anymore, how can I "grab" the vm that was taking care of by pvebld15 from another node like pvebld03.
Is there an "import vm from NAS " or something?

 

[wolfgang]

As I say move the config

mv /etc/pve/nodes/<src_nodename>/qemu-server/<vmid>.conf /etc/pve/nodes/<targest_nodename>/qemu-server/<vmid>.conf

 

[philjans]

So if I understand correctly, if the node is completely gone with no harddrives access possible, there is no way to "import" the VM located on the NAS from a healthy node like you can do in VmWare. You must have at least harddrive access to the defective node to grab the .conf file.
Ok. good to know.

Thank you for your assistance.

 

[philjans]

I also checked my install and see this:

How can I fix it?

 

[wolfgang]

You must have at least harddrive access to the defective node to grab the .conf file.
Ok. good to know.

No, you understand something wrong.

/etc/pve/* is on all cluster members the same.
This is Proxmox Cluster FS and it is replicated over all nodes. So if one node is gone and the VM image is available on the other nodes you can copy the config file on every other cluster member and start the VM.
There is no import necessary.

 

[wolfgang]

How can I fix it?

Install the proxmox-ve package

apt install proxmox-ve

 

[philjans]

Thanks for the clarifications.
I wanted to upgrade but my node pve05 is down now (not pve15 which is the one I already upgraded).


And the restart isn't working:


Where is the issue or where should I look?
tx

 

[wolfgang]

Can you send the pveversion -v of this node?

 

[philjans]

Here you go:

 

[wolfgang]

Please upgrade and reboot

apt update && apt full-upgrade
reboot

 

[philjans]

It worked!
All nodes are on the same level now.
Thank you!