Hello,
We are running Proxmox 5.3 on a cluster (did not had time to upgrade yet this one) and last weekend we had a planned downtime (changed switches). After starting up, nothing was working, the web ui came up, login failed with "No Proxmox VE services running", no VMs were started, cluster quorum was fine but there were some errors related to pinned certificates in pveproxy logs.
Upon investigation i found that the server names were in uppercase (i installed those servers and i specifically remember using lowercase so those were changed runtime somehow, these servers had a few hundred days of uptime), but everywhere else, in /etc/hosts etc, the host names were only in lowercase.
What happened is that in the /etc/pve/nodes directory there were 2 sets of directories created for each server - one in uppercase, one in lowercase. The lowercase ones were the originals containing all the VM configs etc, the uppercase ones were pretty much barebone. I assume this happened on boot time based on the hostnames of the nodes.
After a bit of cold sweat, frantic searching and debugging it turned out the solution was to set the names back to uppercase, reboot and everything works fine.
I don't know if this is still a potential issue but just a FYi if anyone encounters this issue.
We are running Proxmox 5.3 on a cluster (did not had time to upgrade yet this one) and last weekend we had a planned downtime (changed switches). After starting up, nothing was working, the web ui came up, login failed with "No Proxmox VE services running", no VMs were started, cluster quorum was fine but there were some errors related to pinned certificates in pveproxy logs.
Upon investigation i found that the server names were in uppercase (i installed those servers and i specifically remember using lowercase so those were changed runtime somehow, these servers had a few hundred days of uptime), but everywhere else, in /etc/hosts etc, the host names were only in lowercase.
What happened is that in the /etc/pve/nodes directory there were 2 sets of directories created for each server - one in uppercase, one in lowercase. The lowercase ones were the originals containing all the VM configs etc, the uppercase ones were pretty much barebone. I assume this happened on boot time based on the hostnames of the nodes.
After a bit of cold sweat, frantic searching and debugging it turned out the solution was to set the names back to uppercase, reboot and everything works fine.
I don't know if this is still a potential issue but just a FYi if anyone encounters this issue.