Part of this discussion appears in the following thread: Login failed. Please try again I'll try to gather the relevant information here.
I created three new Proxmox VE servers (on older hardware, IBM x3650 M2). Each server in stand alone mode seemed to work just fine. Server names are pve01, pve02 and pve03.
On pve02, I uploaded several ISOs, mainly just testing how the store works, etc. Otherwise, I did nothing else on any of the three other than the initial install.
Using the WebUI on pve01, I created a cluster. As soon as it created, I used the join information to add pve02 and pve03. During the process, I believe it was during the cluster create and the node joins I got the following error in the WebUI: "Permission denied Invalid PVE ticket (401)." However, it appeared each step ultimately worked as the nodes showed up in the cluster list. However, while pve01 had a green checkmark, pve02 and pve03 had grey question marks by their names.
At the time I did this, I had three browser windows open which started to misfire as I clicked around. As I learned from the thread mentioned above, this was most likely because the local logins were now being ignored in favor of the cluster pushed authentication, which, if my assumption in the title is correct, were not getting pushed to the nodes. Thus, the only node I can login on is pve01.
I tried all the things mentioned in the above thread and the main problem indicator is that the /etc/pve directory does not respond properly and seems to hang on many (not all) operations. In fact, /etc/pve/nodes/pve01 is completely non-responsive.
The one additional snafu I noticed, was that pve02 had an invalid entry for itself in /etc/hosts. I corrected it, but that was after adding it to the cluster, so there may be a bad IP in a config file somewhere I can't find.
In theory, there are almost no firewall restrictions on this particular network, so ports 5404 and 5405 should be open. However, I was not able to do anything that validated they where working.
Currently, each server has two NICs. Not knowing better, I foolishly used the "main" one when setting up the Cluster and Nodes.
corosync and pve-cluster appear to be working when querying their status from systemctl. pvecm status appears to look good and shows the cluster as being quorate.
These are development machines and are still being setup. So I am willing to be told I just did something horrifically wrong and need to start over. However, I would like this thread to give me more guidance so if I do start over, I do it better. I would also very much like to get some troubleshooting and/or diagnostic skills so I know a little more about what to test, how to test it, and what to look for. Right now, the logs are just miles and miles of stuff that "looks good to me..."
Thanks much!
I created three new Proxmox VE servers (on older hardware, IBM x3650 M2). Each server in stand alone mode seemed to work just fine. Server names are pve01, pve02 and pve03.
On pve02, I uploaded several ISOs, mainly just testing how the store works, etc. Otherwise, I did nothing else on any of the three other than the initial install.
Using the WebUI on pve01, I created a cluster. As soon as it created, I used the join information to add pve02 and pve03. During the process, I believe it was during the cluster create and the node joins I got the following error in the WebUI: "Permission denied Invalid PVE ticket (401)." However, it appeared each step ultimately worked as the nodes showed up in the cluster list. However, while pve01 had a green checkmark, pve02 and pve03 had grey question marks by their names.
At the time I did this, I had three browser windows open which started to misfire as I clicked around. As I learned from the thread mentioned above, this was most likely because the local logins were now being ignored in favor of the cluster pushed authentication, which, if my assumption in the title is correct, were not getting pushed to the nodes. Thus, the only node I can login on is pve01.
I tried all the things mentioned in the above thread and the main problem indicator is that the /etc/pve directory does not respond properly and seems to hang on many (not all) operations. In fact, /etc/pve/nodes/pve01 is completely non-responsive.
The one additional snafu I noticed, was that pve02 had an invalid entry for itself in /etc/hosts. I corrected it, but that was after adding it to the cluster, so there may be a bad IP in a config file somewhere I can't find.
In theory, there are almost no firewall restrictions on this particular network, so ports 5404 and 5405 should be open. However, I was not able to do anything that validated they where working.
Currently, each server has two NICs. Not knowing better, I foolishly used the "main" one when setting up the Cluster and Nodes.
corosync and pve-cluster appear to be working when querying their status from systemctl. pvecm status appears to look good and shows the cluster as being quorate.
These are development machines and are still being setup. So I am willing to be told I just did something horrifically wrong and need to start over. However, I would like this thread to give me more guidance so if I do start over, I do it better. I would also very much like to get some troubleshooting and/or diagnostic skills so I know a little more about what to test, how to test it, and what to look for. Right now, the logs are just miles and miles of stuff that "looks good to me..."
Thanks much!
