can't connect to host.docker.internal

[dbarrosop]

Description

Not sure what changed between 0.6.2 and 0.6.5 but on <=0.6.2 I could add on my docker-compose file:

        extra_hosts:
            - host.docker.internal:host-gateway

and then run a service on my laptop and connect from the container directly to it via that name. For instance, if I had on my laptop a service running on :8080 from the container I could simply do curl http://host.docker.internal:8080. Now on 0.6.5 I get "connection refused". I noticed that curl http://192.168.5.2:8080 works but I don't think hardcoding IPs is a good solution, specially when sharing the docker-compose file with others.

Version

colima version 0.6.5
git commit: dev

runtime: docker
arch: aarch64
client: v20.10.25
server: v24.0.7
limactl version 0.18.0

qemu-img version 8.1.2
Copyright (c) 2003-2023 Fabrice Bellard and the QEMU Project developers

Operating System

• macOS Intel <= 13 (Ventura)
• macOS Intel >= 14 (Sonoma)
• Apple Silicon <= 13 (Ventura)
• Apple Silicon >= 14 (Sonoma)
• Linux

Output of colima status

colima status
INFO[0000] colima is running using macOS Virtualization.Framework
INFO[0000] arch: aarch64
INFO[0000] runtime: docker
INFO[0000] mountType: virtiofs
INFO[0000] socket: unix:///Users/dbarroso/.colima/default/docker.sock

Reproduction Steps

1. On your laptop run nc -l 8080
2. docker run --rm --add-host=host.docker.internal:host-gateway -it busybox sh -c 'echo asd | nc host.docker.internal 8080 && echo "success" || echo "failed"'
3. docker run --rm --add-host=host.docker.internal:host-gateway -it busybox sh -c 'echo asd | nc 192.168.5.2 8080 && echo "success" || echo "failed"'

For instance:

On one terminal:

$ nc -l 8080

On another:

$ docker run --rm --add-host=host.docker.internal:host-gateway -it busybox sh -c 'echo asd | nc host.docker.internal 8080 && echo "success" || echo "failed"'
failed

$ docker run --rm --add-host=host.docker.internal:host-gateway -it busybox sh -c 'echo asd | nc 192.168.5.2 8080 && echo "success" || echo "failed"'
success

Expected behaviour

Connecting via host.docker.internal should work as it used to on <=0.6.2

Additional context

No response

[abiosoft]

Does it work if you remove the extra_hosts section of your docker compose file?

[dbarrosop]

Yes, it works without it but removing it breaks compatibility with other docker setups. Those lines are there in the docker-compose file to make sure the same docker-compose works across environments. For reference:

moby/moby#40007
docker/for-linux#264

And as mentioned before this was working fine at least up to 0.6.2 (I upgraded this morning from 0.6.2 to 0.6.5 so unsure which version broke this).

BTW, awesome work with colima, huge fan here :)

[abiosoft]

Yeah, I am not asking you to remove it, I am only trying to confirm if it's only the host-gateway that's broken or the DNS.

I suspect it is (docker daemon) configuration related.

Thanks for reporting.

[rfay]

In DDEV we only add that on Linux docker-ce systems, since host.docker.internal is provided by other systems like Docker Desktop/Colima/OrbStack everywhere. However, it would be nice if it worked both ways (provided by the docker setup and added by --add-host=host.docker.internal:host-gateway

[abiosoft]

The issue has been found.

For whatever reason, the docker daemon is not picking up the host-gateway-ip from the daemon config file. However, configuring dockerd via systemd works.

Fixing in #904.

[abiosoft]

This should be fixed by now. You can verify by installing the development version brew install --head colima.

Thanks.

[dbarrosop]

Perfect, I can confirm it works now. Amazing work and thanks!

@rfay FWIW we always add it because it is basically a NOOP in docker-for-mac and other systems that add it automagically and it adds a consistent behavior across development environments.

[skipdegroot-avisi]

@dbarrosop Does this still work for you in 0.6.7? I'm running into the same issue on my Sonoma 14.2.1 machine by the looks of it:

$ colima version && limactl --version && qemu-img --version
colima version 0.6.7
git commit: ba1be00e9aec47f2c1ffdacfb7e428e465f0b58a

runtime: docker
arch: aarch64
client: v24.0.7
server: v24.0.7
limactl version 0.19.1
qemu-img version 8.2.0

$ docker run --rm --add-host=host.docker.internal:host-gateway -it busybox sh -c 'echo asd | nc host.docker.internal 8080 && echo "success" || echo "failed"'   
failed
$ docker run --rm --add-host=host.docker.internal:host-gateway -it busybox sh -c 'echo asd | nc 192.168.5.2 8080 && echo "success" || echo "failed"'
success

I have tried rebooting, starting over after a colima delete and switching from vz to qemu virtualisation. Nothing seems to fix this except removing the host.docker.internal:host-gateway line which also isn't preferred here due to multiple different development environments within the same team.

[dbarrosop]

Yeah, this seems broken again in 0.6.7

[markaschneider]

Yup, not working for me in 0.6.7

[opticoder]

Not working with 0.7.0 also

[abiosoft]

Can you try the current development version? brew install -head colima.

[opticoder]

Can you try the current development version? brew install -head colima.

Just checked - works fine. Hope the fix will land in the next colima version