Networking issue after 0.70 -> 0.7.1

[ragboy]

Description

After upgrading from 0.7.0 to 0.7.1 I noticed that I could not login to my home gitlab registry. I host gitlab in a kubernetes cluster on my home network. The subnet of my home net is 10.0.2.0/24. My mac is on 10.0.2.99, and the k8s system is in that subnet. When I try to login I get "connection refused". So if I ssh into that colima instance, I can try using curl, and it works outside of my net, but to anything inside my subnet, I get "connection refused". I even tried using curl and telnet to various services on my subnet and they all refuse connection. I manually downgraded to 0.7.0, and then all is well.

Version

This is downgraded:
colima version v0.7.0
git commit: 4b14e8a

runtime: docker
arch: aarch64
client: v27.1.1
server: v27.1.1
limactl version 0.22.0
qemu-img version 9.0.2
Copyright (c) 2003-2024 Fabrice Bellard and the QEMU Project developers

When I use brew for colima to 0.7.1, then it fails.

Operating System

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

Output of colima status

INFO[0000] colima is running using QEMU
INFO[0000] arch: aarch64
INFO[0000] runtime: docker
INFO[0000] mountType: sshfs
INFO[0000] socket: unix:///Users/rgarcia/.colima/default/docker.sock

Reproduction Steps

1. Use 0.7.1
2. ssh into colima and curl something on the local subnet.

Expected behaviour

successful connection, like with 0.7.0

Additional context

No response

[abiosoft]

A feature that was added to support host IP addresses for docker port binding. That is most likely the cause.

A fix would be to hide the feature behind a flag and have it disabled by default.

Even though you probably do not need this, just to validate my hypothesis, can you start with --network-address flag and see if the behaviour stays the same?

Thanks.

[ragboy]

I will check, and let you know.

[ragboy]

with 0.7.1 w/o --network-address fails:

rgarcia@colima:/Users/rgarcia$ curl -Iv https://xxxxxxxx.com
* Host wake9.com:443 was resolved.
* IPv6: (none)
* IPv4: 10.0.2.62, 10.0.2.63, 10.0.2.61
*   Trying 10.0.2.62:443...
* connect to 10.0.2.62 port 443 from 10.0.2.130 port 56418 failed: Connection refused
*   Trying 10.0.2.63:443...
* connect to 10.0.2.63 port 443 from 10.0.2.130 port 41818 failed: Connection refused
*   Trying 10.0.2.61:443...
* connect to 10.0.2.61 port 443 from 10.0.2.130 port 34476 failed: Connection refused
* Failed to connect to wake9.com port 443 after 7 ms: Couldn't connect to server
* Closing connection
curl: (7) Failed to connect to wake9.com port 443 after 7 ms: Couldn't connect to server

and then when I add --network-address, it works fine:

rgarcia@colima:/Users/rgarcia$ curl -I https://xxxxxxxx.com
HTTP/2 200
date: Mon, 12 Aug 2024 14:13:35 GMT
content-type: text/html; charset=UTF-8
x-pingback: https://wake9.com/xmlrpc.php
link: <https://wake9.com/wp-json/>; rel="https://api.w.org/", <https://wp.me/P9ATpQ-D>; rel=shortlink
vary: User-Agent
strict-transport-security: max-age=31536000; includeSubDomains

[abiosoft]

Thanks for confirming, a fix would be out soon.