It’s because of that decision that I decided to maintain it separately.
It’s important to recognise that the TorVM is compatible with Qubes
firewalling, unlike Whonix. It’s lightweight, and serves one purpose
providing a Tor Proxy.
It’s for users to decide if this serves their needs.
I never presume to speak for the Qubes team.
When I comment in the Forum I speak for myself.
Of course.
Clone a debian minimal template - say template-tor
In the template, add the 3isec repository definition in
/etc/apt/sources.list.d/3isec.list:
“deb [signed-by=/usr/share/keyrings/unman-keyring.gpg] Index of /4.3 trixie main”
Get a copy of my signing key, and install it in /usr/share/keyrings/unman-keyring.gpg
In the template, install the 3isec-tor package - sudo apt install 3isec-tor
Create a qube using template-tor
Set preferences to provides_network True qvm-service QUBES 3isec-tor on
Start the tor qube.
Any qubes with netvm set to the tor qube will access over Tor.
You can run nyx to monitor bandwidth and circuits, and control the tor
daemon.
I never presume to speak for the Qubes team.
When I comment in the Forum I speak for myself.
Should have mentioned updates:
I run cacher behind tor, and have done so for years. I update dom0
behind tor qube.
I rarely have user reports of problems. Fewer than I seem to read of
Whonix.
I never presume to speak for the Qubes team.
When I comment in the Forum I speak for myself.
The Tor daemon is not running, likely because you have not enabled the
service.
In the tor qube, run systemctl status 3isec-tor, and systemctl start 3isec-tor. This will help identify the issue.
I never presume to speak for the Qubes team.
When I comment in the Forum I speak for myself.
After enabling and starting it I can’t connect to any host - neither from the AppVM, nor from the TorVM gateway itself. I triple checked that the service is running. How to proceed?
BTW, isn’t it a problem that with non-running service it falls back to clearnet, thus effectively deanonymizing all client qubes? IMO, the correct behaviour should be to drop all instead. What do you think?
TorVM is a lightweight ProxyVM, not a drop-in replacement for Whonix Gateway, and the same logic applies to core-tor in Liteqube. The correct behaviour is to never carry expectations from one stack to another with different maintainers and use cases.
The correct behaviour is to use Tor when the service is running, otherwise do not use Tor when the service has stopped. The category error is treating everything outside of the service as dependent on your specific network topology, threat model, and use case(s), when the TorVM service already works based on its design constraints in isolation.
Considering the goal of TorVM is anonymity, I don’t see how it can be correct to allow clearnet connections at all - that’s simply contrary to the whole idea.
In any case, even with a running service I can’t connect to any host, as mentioned, so that is the main issue for the moment.