HELP!? Lost container configuration files, how can I regenerate/restore them?

[oeginc]

Ok, first - before I get chewed out for doing this, I know what I did was wrong and I will not doing it again.. At least by accident.

So what I did was rebuilt all of my machines destroying my old cluster, migrating machines to new hosts as needed to reformat and reinstall fresh on the other hosts.

Once I was done, I started creating a new cluster. I went to add my second machine (I have 8 total right now) but it already had VM's running and complained.

I checked the documentation and it said something like: Make sure you don't have any VM's running.. blah blah... Something or another... And then it said something about it could cause a conflict in ID's.

Well, *all* of my ID's are unique across *all* of my hosts, so I figured that wouldn't be a problem for me. So I -force'd the node to get added to the cluster and then, *BAM* noticed that *ALL* of my host configuration files were magically gone (which in the big scheme of things makes sense).

Why did I do this? Well, to be honest - it's a production environment and I couldn't afford to take the time to migrate or backup/restore all of the hosts (about 18-25 of them) on every server I needed to add to the cluster. As long as my ID's don't conflict, I don't understand why that's necessary...

Anyways.. The containers appear to still be running, but I can't do anything with them because pvectl wants the /etc/pve/id.conf files to exist first.

I have some older backups of most of the containers, there has to be some way of generating these config files from the backups without actually restoring every backup - right? How/where does vzdump and vzrestore keep the container information?

Also, on a side note... Would this have worked okay if I backed up my /etc/pve/*.conf files and then restored them to /etc/pve/nodes/<node> or whatever after adding to the cluster?

Thanks in advanced!

 

[wahmed]

I feel your pain . Knowledge gained from mistakes are the best ones. But painful to learn them nonetheless.

Once I was done, I started creating a new cluster. I went to add my second machine (I have 8 total right now) but it already had VM's running and complained.

So do you have 2 clusters right now? And all your VMs are still on the old cluster?

 

[oeginc]

No, my old cluster is gone (I have physically reformatted all of the boxes).

So... After a little experimenting and reading thru the PVE libraries I found that the VPS configuration file *IS* stored in the backup archive (as I had expected) in /etc/vzdump/vps.conf

So I am going thru each backup, extracting ./etc/vzdump/vps.conf and then moving that file to /etc/pve/openvz/<CTID>.conf

It *seems* to be working so far, but I'm not sure if there are any long-term "bad things" gonna happen...

I have also noticed that there seems to be a backup or two missing (not sure why), is there some way to recreate the container configuration from a running container, or do I just have to do it manually and hope I guess the parameters correctly?

 

[wahmed]

I am not too familiar with Container so cannot comment on that. I am sure there is a command for CLI to see status of running Container.

Just out of curiosity, when you were killing old cluster, where were your VMs?

 

[oeginc]

As I mentioned originally, I migrated them off to the other nodes and kept moving them around until I had everything reformatted and reinstalled. It worked just fine (you can migrate to nodes that aren't in your cluster) - it just took a very long time to complete and I was hoping not to have to do that again to create the new cluster.