This message was deleted.
# rancher-desktop
a
This message was deleted.
f
Did you enabled "Administrative Access"? Rancher Desktop can only get an external IP address when running the socket_vmnet daemon as
root
. It looks like the settings validation code allows you to enabled socket_vmnet without admin access, which will not work
m
No, I did not I’ll check it right away. Thank you
Hmm, it still does not work…
Copy code
❯ kubectl get node/lima-rancher-desktop -o jsonpath='{.status.addresses[?(@.type=="ExternalIP")].address}'
192.168.2.109
❯ nc -zv 192.168.2.109 80                                                                                                                                                                                                    
^C
packets seem to be dropped. but:
Copy code
❯ rdctl shell
lima-rancher-desktop:/Users/seyhbold$ nc -zv 192.168.2.109 80
192.168.2.109 (192.168.2.109:80) open
f
I've meetings coming up now; will look at it some more later, as something seems wrong on my machine too
m
Thank you so much.
f
I'm still looking into it; I could reproduce the issue, but it was fixed after a reboot. I have another machine, that also has the problem right now, so I'm keeping it running as-is, but I haven't been able to figure out why it sometimes works, and sometimes doesn't.
m
Thank you. I have the problem even after a reboot.
Hej @fast-garage-66093 did you have the opportunity to look into this topic?
f
Sorry, I had to work on something else in the meantime. I think we should create a Github issue for it, and see if somebody else can take a look. I hope to be able to get back to it later this week.
m
Sure, thanks
f
I've finally figured it out; took me almost 2 days of work... First I had to bisect the Rancher Desktop commits between 1.10,0 and 1.11.0 to find out which commit broke things. Turns out that it was a bump of Lima. So I had to do the bisect again for Lima between 0.16 and 0.18. Found the commit as well, but could still not repro the issue with Lima by itself.
It turns out to be that when you create the VM with Rancher Desktop it initially doesn't have the bridged interface. During first boot, the
/etc
directory is copied to a persistent volume (
/mnt/data/etc
) so changes persist between reboots, and the
/etc
directory is replaced by a symlink to
/mnt/data/etc
because the original
/etc
is in a RAM disk (tmpfs).
Now when you allow admin access, the bridged interface is added and correctly configured during boot. But this configuration updates the
/etc/network/interfaces
in the RAM disk and not on the volume. But it doesn't matter as the networking service is started before
/etc
is replaced by the symlink.
Now the change in Lima was adding a restart of the networking service, but this restart was happening after the directory has been replaced by the symlink. So it was using the original network configuration from the first boot, which doesn't include the bridged network...
This is probably more than you wanted to know; I just wanted to explain why it was a bit tricky to figure out.
You can work around this problem by running a few commands inside `rdctl shell sudo -i`:
Copy code
mkdir /bootfs
mount --bind / /bootfs
cp /bootfs/etc/network/interfaces /etc/network/
Then reboot Rancher Desktop and the Traefik should now be accessible on the bridged network.
Let me know if it works for you. If you ever do a factory-reset then you'll have to redo this change
This issue is fixed in the patch release: Release Rancher Desktop 1.11.1
m
Oh wow, thank you so much
I just came to test this today. It is indeed working now. Even without any resets. The update fixed this immediately. Thank you