policies seem to have no effect

[amjelnes]

Describe the problem

Policies seem to have no effect when limiting access on different ports and protocols

To Reproduce

Steps to reproduce the behavior:
Create policies fx. group clients -> Exit node UDP/53
Disable all other policies and ping a device on exit nodes network, ping will respond from the device, likewise external websites can still be browsed, even though no other traffic than UDP/53 should be allowed.

Expected behavior

Would expect that ping would not respond and that no websites could be browsed

Are you using NetBird Cloud?

Self hosted

NetBird version

latest per 2nd may 2025 16:16 CEST

Is any other VPN software installed?
no

Debug output

To help us resolve the problem, please attach the following debug output

netbird status -dA

As well as the file created by

netbird debug for 1m -AS

We advise reviewing the anonymized output for any remaining personal information.

Screenshots

If applicable, add screenshots to help explain your problem.

Additional context

Add any other context about the problem here.

Have you tried these troubleshooting steps?

• [x ] Checked for newer NetBird versions
• [x ] Searched for similar issues on GitHub (including closed ones)
• [ x] Restarted the NetBird client
• [ x] Disabled other VPN software
• [x ] Checked firewall settings

nbstatus.txt
netbird.debug.2759613363.zip

[lixmal]

Have you assigned an access control group to the exit node route in network routes? Otherwise, all access will be allowed by default.
This group can be distinct from the group the exit node peer is in.

[amjelnes]

Have you assigned an access control group to the exit node route in network routes? Otherwise, all access will be allowed by default. This group can be distinct from the group the exit node peer is in.
No wasn't aware of that, done now, but I'm confused:
I have 1 exit node this is the only member of a group called exit
I have a number of clients these are only member of a group called clients

I have now added client group to access control groups in exit node route in network routes, also client group is added to distribution groups in the same network route .

I can still ping the local device even though only UDP/53 is active.

If I then enable 80,443,53 I can't browse websites, but I can still ping the local device.

If I then add Exit group to Access control policies Groups in the network routes for the exit node I can browse and ping (ping policy disabled, only ports 80,443,53 enabled)

If I remove Exit group from access control policy and add it to distribution groups I can't browse websites but can still ping.

In other words pinging local device seems indifferent to policies

[lixmal]

It's not quite clear to me.

Just to clarify, the setup should be roughly:

Clients group contains your windows client only
EXIT contains only the exit node peer

In Network routes:

Distribution group is Clients
Access control group is EXIT

Policies are fine so far.

In other words pinging local device seems indifferent to policies

What IP is that? Do you have any routes on the client that bypass the exit node route?

Can you create a debug bundle on the exit node?