[SOLVED] /etc/pve folder is empty

[sdssdice]

Hi all,

today i restarted one node of our proxmox cluster with pve 7.1-10. This node now has a empty /etc/pve directory. I dont see any reason for that. So my questions are:

1. Is it possible to debug / find the issue for that?
2. There is / was a VM with ID 100 with local storage on this node. Now i would need to remove this VM but how should i do that without having this node online in the cluster? (The other nodes still show the VM with a question mark in the GUI)

After that i would re-install this single node and bring it back to the cluster.

 

[Matthias.]

Seems like the node can't connect to the cluster.
Is there any information in the log files? What's the output of `pvecm status`?
Did you change anything before restarting?

 

[sdssdice]

Hi Matthias,

thanks for your answer. I checked ping and SSH connection but after more debugging for the service pve-cluster, i saw that it was not able to find his own IP address, because the hostname in the hosts file has been added to 127.0.0.1 and to the ip address. The error was:

pmxcfs[1680672]: [main] crit: Unable to get local IP address

So after removing the entry for 127.0.0.1 and restarting pve-cluster service, the /etc/pve directory appears again.

I didn't know that this directory is mounted from this service. Can you confirm this?

Also, is it recommended to have the FQDN (like proxmox01.domain.com) in the /etc/hostname file or jus the short name (like proxmox01). In the hosts file, both is for the ip address.

 

[Matthias.]

Yes, exactly, it's mounted. Our docs describe how that works: https://pve.proxmox.com/pve-docs/pve-admin-guide.html#chapter_pmxcfs.

I think the short name, no?

 

[imagina]

For later:

I was able to mount /etc/pve chrooting to the environment from rescue mode and running:

pmxcfs -d -l