Need more doc for libvirt-console-proxy

[Guy Godfroy]

Hello,

I'm making a web app for my company that will enable different teams to
manage their own VMs. I wish to make possible to interact with each VM
console, so I plan to use some xterm.js with websockets.

So I discovered libvirt-console-proxy [1] when I looked for something to
put a libvirt console into a websocket. That seems like the right tool
for the job.

The only doc I found is this article from 2017 [2]. After trying to
understand from this article and from --help, I still have many
questions. I am really bad at reading code so I can't even get answers
from the sources.

My main concern is: How a client is supposed to talk to the proxy? It is
said that a security token must be provided. How? HTTP header? Which
header? Am I missing something in websocket protocol? I think an example
client implementation would help a lot.

Also, I tried to use virtconsoleresolveradm to set up metadata on my
domains like explained in the article [1] :

./virtconsoleresolveradm enable milou
Enabled access to domain 'milou'

But that doesn't seem to do anything (except defining the metadata
namespace in the XML):

    virsh metadata milou http://libvirt.org/schemas/console-proxy/1.0
    <consoles/>

I precise that I have already this in my XML:

    <serial type='pty'>
      <target type='isa-serial' port='0'>
        <model name='isa-serial'/>
      </target>
    </serial>
    <console type='pty'>
      <target type='serial' port='0'/>
    </console>

Should I remove that? Should I edit that?

Thanks for your help.

Guy Godfroy

[1] https://gitlab.com/libvirt/libvirt-console-proxy
[2]
https://www.berrange.com/posts/2017/01/26/announce-new-libvirt-console-proxy-project/