[SOLVED] Invalid PVE ticket (again)

[ozdjh]

I've tried to get help with this via a support ticket but the responsiveness has been absolutely terrible and the information unhelpful.

So, I'll ask here. We are seeing the dreaded "Invalid PVE ticket (401)" yet again. It happens every couple of months for no reason we can find. Restarting pvedaemon and pveproxy usually fixes it but it hasn't this time. It's not a browser issue as we are also seeing problems with WHMCS talking to the cluster at the same time we can't use the web UI.

So, regarding the usual questions :

1. Yes, the clocks are in sync. Each node is running NTP with several stratum 1, 2 and 3 peers
2. Yes, the cluster appears healthy. "pvecm status" shows no problems
3. Yes, the networks are reachable and we haven't suddenly added a firewall rule to break the cluster networks
4. Yes, I've restarted pvedaemon and pveproxy, but why should I need to anyway.

There was one Cert that said it expired last February so I've recreated it but can't see how that would be the problem otherwise why does it work most of the time and why didn't it just totally break in February?

Can those of you who understand how this ticket system works provide some insight into what other than time mismatch and network connectivity problems can cause this problem? For 2 days now we've had intermittent problems managing our cluster and our customers managing their VMs. I'd appreciate any pointers.

Thanks
David

 

[tom]

I've tried to get help with this via a support ticket but the responsiveness has been absolutely terrible and the information unhelpful.

You talk about a support ticket from your third party toolkit?

I cannot think of such a bad experience with our Proxmox support.

 

[ozdjh]

No, your proxmox team sadly.

Anyway, got any suggestions or some education as to how the ticket auth works so we can try to figure this out?

 

[tom]

I just checked your ticket.

Seems that you do not have issues with Proxmox VE, you have issues with invalid tickets via your third party WHMCS plugin from modulesgarden.

=> Please contact Modulesgarden, I am pretty sure they can assist.

 

[ozdjh]

Tom, no you are wrong, please re-read the ticket. If we are logged into the Web UI we get errors in the VM status pane etc. Eventually we just get kicked out of the Web UI and we have to try to log in again, which may or may not work. So it's a PVE problem.

At the same time we are seeing those problems, we also cannot access the cluster from WHMCS. So the API just breaks from everywhere.

 

[tom]

=> if you think its a Proxmox issue, please continue on https://my.proxmox.com/

 

[ModulesGarden]

I just checked your ticket.

Seems that you do not have issues with Proxmox VE, you have issues with invalid tickets via your third party WHMCS plugin from modulesgarden.

=> Please contact Modulesgarden, I am pretty sure they can assist.

@ozdjh If some feature doesn't seem to be working as it should then kindly contact us so your problem can be looked into.

 

[ozdjh]

Just to close this thread out, after an unfortunately slow start, the Proxmox support team have been working well on this problem. It appears to be a bug regarding a rare race condition in the rotation of the auth keys. There's discussion of a patch on the dev mailing list. Let's hope this reduces the occurrence of the Invalid Ticket error.

 

[ozdjh]

And a final comment, this update is now available on the enterprise repo as libpve-access-control_7.2-4