Incus web UI doesn't respect core.https_address after reboot

Incus
1

I’m not sure if this is a bug with Incus or I’m just doing something wrong.

After rebooting my Incus host, I’m not able to access the web UI anymore.

I can fix this by setting core.https_address to something different and then setting it back to the original value.

For example: incus config set core.https_address '10.8.8.2:8001'
And then: incus config set core.https_address '10.8.8.2:8000' (I want to listen on 8000)

Doing this seems to force incus to start listening on the correct address and port.

To give some more info, running ss -lnt4Q before flipping the setting shows the following:

State                Local Address:Port
LISTEN               *:32803
LISTEN               0.0.0.0:22

After setting core.https_address:

State                Local Address:Port
LISTEN               10.8.8.2:8000
LISTEN               0.0.0.0:22

Has anyone else run into this?

2

Likely Incus started before that IP address was available on the system.

1 Like
3

Ah yeah I should have checked the log.

incus incusd[1428]: time="2025-08-03T18:13:10-05:00" level=error msg="Cannot currently listen on https socket, re-trying once in 30s..." err="Bind network address: listen tcp 10.8.8.2:8000: bind: cannot assign requested address"
incus incusd[1428]: time="2025-08-03T18:13:41-05:00" level=error msg="Still unable to listen on https socket" err="Cannot listen on network HTTPS socket \"10.8.8.2:8000\": listen tcp 10.8.8.2:8000: bind: cannot assign requested address"

I run OPNsense in a VM (via Incus) which is my only router/gateway, and the interface I want Incus to listen on receives its IP via DHCP. So I think what’s going on is that OPNsense takes long enough to start up that the interface doesn’t receive its IP in time.

Two solutions:

  1. Give the interface a static IP
  2. Create a oneshot systemd service that waits for the interface to get its IP and then run the incus config set commands
1 Like