Error "Connection error 501: Not Implemented"

[lcasale]

I'm getting a "Connection error 501: Not Implemented" error for some pieces of functionality. For example, I'm unable to remove or edit the hardware for VM's. I currently have Pound configured as a reverse proxy to access Proxmox outside of my network. If I access Proxmox locally I don't have any issues. Anyone know why Pound is breaking some pieces of Proxmox?

FYI...
Pound is listening on ports 80 and 443. The backend Pound configuration for the Proxmox server is port 8006. Apache on the Proxmox server is configured to force SSL.

 

[Tozz]

It looks to me you should ask this in a Pound support forum, not Proxmox. But this setup isn't gonna work even if you manage to configure Pound properly. The VNC console uses a separate port (500x where X depends on the current active console sessions) outside of Pound and/or the Proxmox webserver. So you will never be able to use VNC console in this setup.

Anyway, the "501 Not Implemented" means that your client is doing a HTTP request which Pound does not understand. Pound probably only understands HEAD, GET and POST. Proxmox seems to be using some other request types. You need to enable those in Pound.

If you Google for "pound 501 not implemented" the first hit has your answer:

http://www.mikedellanoce.com/2011/09/enabling-pound-proxy-support-for-http.html

 

[lcasale]

That fixed the problem, thanks Tozz! VNC is working as well.

 

[Retheesh]

Hello Lcasale,

I too am facing the same issue. My proxmox servers are behind a reverse proxy server, able to connect and also perform the VNC sessions. However i am getting the same error "Connection error 501: Not Implemented" when i try to remove any instance.

Can you please let me know what is it that you did to make this work?

Regards,
Retheesh

 

[lcasale]

Hi Retheesh, I had to modify the xHTTP value in the POUND settings to 2 or higher.

http://linux.die.net/man/8/pound

 

[Retheesh]

Hi lcasale,

Thank you, Your suggestion worked like a charm. To resolve the same problem on Zen LB, set the HTTP verbs accepted to "+standard WebDAV Verbs"

Cheers,
Retheesh

 

[OdoMarTus]

Hi.
I have the same problem, but I use CloudFlare proxy.
Does anyone know the solution?
I'll just mention that the problem appeared only this weekend.
Everything was working fineIt's good that everything works fine in the local network.

 

[itsthejb]

I appear to now be getting the same issue as @OdoMarTus. I'm also serving the web UI through a Cloudflare tunnel. However, possibly also significant is I also just updated my machine to macOS Sequoia... Otherwise I can connect from the local ip on port 8006 as before

 

[tripheo2410]

Hi, I also have the same problem as @OdoMarTus and @itsthejb, running web UI through Cloudflare tunnel via port 8006.

 

[itsthejb]

@OdoMarTus @tripheo2410 I don't see any evidence that any upgrade has happened on the Proxmox side that would be causing this (checked the apt log). Therefore this might not be the right place to discuss the prob - this may be something cloudflare-related instead?

In any case, I haven't investigated very deeply yet, but certainly would like to get back to the previous setup since I now don't have any easy way to admin when away from home

 

[itsthejb]

I found the following existing post on CF forums: https://community.cloudflare.com/t/connection-failed-error-501/723372/2. It's not the most descriptive post... Perhaps you can all cross-post there to get visibility? Alternatively it may be helpful to make a more descriptive post there

 

[OdoMarTus]

Hi - This CF entry is mine ;]

 

[itsthejb]

Haha, ok thanks @OdoMarTus. Perhaps it would help to add more context, that it is affecting the Proxmox Web UI? I haven't seen their issue form before, but it seems rather thin on info to me

 

[OdoMarTus]

Hi.
You've put me on a different track.
I use a VM with OpenWRT as a router and I've launched the CloudFlare agent on it.
And when it's used, the above problem appears.
When I launch the VM with the CloudFlare agent, everything works fine.
The problem lies with bugs in the CloudFlare agent on OpenWRT - or rather in OpenWRT.

Do you also use routers with OpenWRT?

 

[itsthejb]

I suspect the culprit is the `cloudflared` release 2024.10.0 from a few days ago: https://github.com/cloudflare/cloudflared/releases/tag/2024.10.0. I've reported it on their Github https://github.com/cloudflare/cloudflared/issues/1337

Note that you can click the `>_ Shell` disclosure and select `noVNC` and this implementation is still working

 

[OdoMarTus]

I suspect the culprit is the `cloudflared` release 2024.10.0 from a few days ago: https://github.com/cloudflare/cloudflared/releases/tag/2024.10.0. I've reported it on their Github https://github.com/cloudflare/cloudflared/issues/1337

Note that you can click the `>_ Shell` disclosure and select `noVNC` and this implementation is still working

Hi.
Thank you very much, because I didn't know exactly where and how to report the problem.

I've had the problem for about a month, not days.

I'll add my info.

 

[itsthejb]

I'm now working around the issue by reverting to a backup of the CT that had the working 2024.9.1, and then holding the package to prevent upgrade for the time being.

Looking at the size of CF's Github issues list doesn't fill me with confidence this will get fixed anytime soon, though....