CTs are no longer shown after cluster creation

[Erico]

Hello,

I have 6 CTs on node 192.168.15.23.
Now, I'll insert two other nodes, and I decided to create a cluster.
Following Proxmox_VE_2.0_Cluster, after execute suggested commands on 192.168.15.23:

# pvecm create modcs
# pvecm add 192.168.15.23 -force

CTs are no longer shown: with directory /etc/vz/conf/ empty, nevertheless /var/lib/vz/private/ contains CTs data.
Googling about it results on sites, like this, that involves delnode parameters:

pvecm delnode <Node>

I'm not sure to execute commands that involves delnode parameter to recover what is not been shown.

After reading more carefully Proxmox_VE_2.0_Cluster, I noted a very important recommendation: "Please note, the node cannot hold any any VM´s.". However, there are no warning about CTs, or CTs erasing...

During insertion of 192.168.15.23 on cluster, there is a message notifying backup of old data.

So, does anybody know:
- why CTs are no longer shown?
- how can I do to display them again?
- mainly, where are configuration files of /etc/vz/conf/ directory?
- it'll be necessary to perform any restore operation? If yes, any recommendation?

Thank you so much in advance,

Erico.

 

[wahmed]

From your posting it is not clear if you already had a cluster on 1 node. If you did not, then you do not have to use -force option while adding a node to a cluster. On a cluster all your configuration files and cluster settings are in /etc/pve. This is the folder that gets synced to all other nodes in the same cluster. You will most likely find your CT configuration in /etc/pve/openvz/. You may have to copy all the CT configuration files from /var/lib/vz/private/ to /etc/pve/openvz to see them from Proxmox GUI.

 

[Mr.Holmes]

Hello,

# pvecm create modcs
# pvecm add 192.168.15.23 -force

The mistake was "pvecm add" - this command has to be processed only in the additional nodes in the cluster (and you must add them before a VM or CT was created there) - in the first one "create" is enough (and existing CTs/VMs will remain).

 

[Erico]

From your posting it is not clear if you already had a cluster on 1 node.

No, I don't. I was trying it for the first time.

If you did not, then you do not have to use -force option while adding a node to a cluster.

I tried to do it without -force, without success.

You will most likely find your CT configuration in /etc/pve/openvz/.

They aren't there:

root@ifal-cloud-pm2:~# updatedb
root@ifal-cloud-pm2:~# locate 100.conf
root@ifal-cloud-pm2:~# locate 101.conf
root@ifal-cloud-pm2:~#

Really, they aren't anywhere. It was not possible to locate them.
Further comments will be extremelly helpful.

There is any way to rollback cluster creation and recover previous state?
I noticed that, while cluster creation, pvecm command reported a message about perform a backup of old configuration.

Thanks a lot,

Erico.

 

[Erico]

Thanks by your comment.

Problems are still remain. Any further recommendations?

Thanks again,

Erico.

 

[Mr.Holmes]

Problems are still remain. Any further recommendations?

Hello Erico,

With "Problems" you mean that you cannot restore the CTs? I don´t know how it works in detail, but I think in the moment of "pvecm add" the filesystem which contains /etc/pve/ is reinitialized and all data which were stored there before are lost. On the other hand: only the configuration files are there, the data are in the datastore you specified. That means, if you recreate /etc/pve/openvz/<ct-id>.conf manually it should work. This is not too complicate, the structure of that kind of file is easy to understand when you create an example.

Maybe somebody else has a better idea and knows a way to restore the lost data directly....

kind regards

Mr. Holmes

 

[Erico]

Thank you for your feedback Mr Holmes.

With "Problems" you mean that you cannot restore the CTs?

I had failed because proxmox GUI wasn't shown the templates. After read proxmox gui don't show files on storage, and perform downgrade of libpve-access-control:

apt-get libpve-access-control=3.0-6

it was possible to create a new CT and cp configuration files, as suggested by you. This approach works fine.