[SOLVED] Removed from Cluster but already in list via Web

[PaoloHolzl]

I hade a machine connected to the Cluster, but no more functioning (for a hardware problem).
So I hade remove it from cluster, stopping cluster configuration on it.
Now the machine is starting 'alone' and VM are functioning.

The problem is that is remaining in Web interface list of the Cluster with red cross icon.
I have also into Cluster shell removed directory /etc/pve/nodes/PCREMOVED is remaining.
pvecm nodes show only all correct functioning nodes.
If I go at top of web menu tree of cluster at right I obtain '/etc/pve/nodes/PCREMOVED/pve-ssl.pem' does not exist! (500).
Howho I can remove PCREMOVED from cluster and prof web interface and resolve error?

 

[hd--]

Hello !

First of all, for this use case there is an article in the Documentation, next time i'd recommend following these steps.
Secondly, could you please describe step by step what you've done, as i cannot totally follow your description ?

 

[PaoloHolzl]

Hello !

First of all, for this use case there is an article in the Documentation, next time i'd recommend following these steps.
Secondly, could you please describe step by step what you've done, as i cannot totally follow your description ?

Ok I try to explain better.

On a machine of the cluster:

pvecm nodes

Membership information
----------------------
    Nodeid      Votes Name
         1          1 ProxMox01 (local)
         3          1 ProxMox03
         4          1 ProxMox10

Node ProxMox02 is not present and will remain out of the cluster because no more existent.

Directory ..nodes/ProxMox02 was removed
When I select Datacenter in a server into the cluster this is the response:



Howto remove ProxMox02 from the list and also the error?
Thank you

 

[hd--]

In the documentation, i already posted

Documentation,

it says:

As the configuration files from the other nodes are still in the clusterfile system, you may want to clean those up too. After making absolutely sure that you have the correct node name, you can simply remove the entiredirectory recursively from /etc/pve/nodes/NODENAME.

The node’s SSH keys will remain in the authorized_key file. Thismeans that the nodes can still connect to each other with public key authentication. You should fix this by removing the respective keys from the/etc/pve/priv/authorized_keys file.

have you already tried this ?

 

[PaoloHolzl]

In the documentation, i already posted

it says:

have you already tried this ?

I stopped when

pvecm nodes

didn't show ProxMox02.
Now I have try

pvecm delnode ProxMox02

even without seeing it
and this solved.
Thank's a lot

 

[no-usernames-left]

@hd-- I just went through this myself, and the documentation is actually missing a crucial step for two-node clusters.

pvecm delnode NODENAME complained about a lack of quorum and refused to delete the node. As a result, I had to pvecm expected 1, which I only found mentioned lower down in a separate section (for separating a node without reinstalling, which was not the case).

After adjusting the expected count, the node could be deleted successfully. It still remained in the sidebar, but removing the entries referring to it in /etc/pve/priv/known_hosts and /etc/pve/priv/authorized_keys caused it to disappear.

I was entirely unaware of the step rm -rf /etc/pve/nodes/NODENAME until reading this thread, as it is also not in the section for removing a node, but only for separating without reinstalling.

tl;dr: the documentation section covering removing a cluster a node should be revised to add mention of adjusting expected node count for quorum for two-node clusters, as well as cleanup of known_hosts, authorized_keys, and /etc/pve/nodes/NODENAME.