Podman pull does not work with IPv6

[archneon]

Issue Description

Hi there,

VERSION:
podman version
Client: Podman Engine
Version: 4.9.3
API Version: 4.9.3
Go Version: go1.22.2
Built: Thu Jan 1 00:00:00 1970
OS/Arch: linux/amd64

PROBLEM:
I have VPS (Ubuntu 24.04) on OVHCloud. When you start a new server they are now automatically adding IPv4 + IPv6.
The issue is that i have noticed that the command:
podman pull docker.io/postgres:17-alpine
Trying to pull docker.io/library/postgres:17-alpine...
basically it...was pulling forever....
I retried several times, but it didn't work.

After a while i finally figured out that this is IPv6 issue. I temporary disabled IPv6 to confirm this:
sudo sysctl -w net.ipv6.conf.all.disable_ipv6=1
sudo sysctl -w net.ipv6.conf.default.disable_ipv6=1

And yes, podman pull works now:
podman pull docker.io/postgres:17-alpine
Trying to pull docker.io/library/postgres:17-alpine...
Getting image source signatures
Copying blob 6eecb7b8b1c2 done |
Copying blob f18232174bc9 done |
Copying blob 7f99212d21e8 done |
Copying blob b323a81400d6 done |
Copying blob bed058a4433b done |
Copying blob 2b8f2a5cb6de done |
Copying blob 1badcef40f1c done |
Copying blob e835d685be33 done |
Copying blob ec828f5c7dec done |
Copying blob efee735d65f7 done |
Copying config 2911229feb done |
Writing manifest to image destination
2911229feb1ea901ecd374add415d146c731407f0b3129f0a5c6082ed7c0c0ec

I am wondering, what is the best practice in this case? Should i just completely disable IPv6 on my VPS or is there a better solution / best practice for this issue?

Thanks in advance,
Jim

[baude]

can you please reproduce this with the latest podman version 5.4 or main? Also, I dont think this is a case of podman cannot do this with ipv6 because I know some of our developers are on Ipv6 by default. are you doing anything like podman running in a container ?

[baude]

@Luap99 anything to add here?

[Luap99]

Ipv6 connections should work fine, the pull code uses the normal golang std http client AFAIK and I am very sure that should works with ipv6.

Without specifics it is impossible to tell, at the very least --log-level debug might help to show which requests are being made (and pipe the output to some file to disable the interactive progress bars which reset the terminal output constantly).
strace -f would also be helpful to see the socket connection read/writes.

Directly making ipv6 request against docker.io with curl or something to verify the it works might also be a good idea.

And of course please test with the latest podman version.

[archneon]

Thanks for reply guys! I did a test pull for the same image with Docker. Same thing happened. Here is the reason why this is happening on IPv6.
https://serverfault.com/questions/1167925/docker-429-too-many-requests

Hopefully this issue willl get resolved.

kind regards!

[Luap99]

If you are hitting the rate limit there should be an error returned saying so, if it pull forever that would not match the the definition of a rate limit AFAICT.

Obviously if you hit that rate limit there is nothing podman can do about it, the limits are set by docker https://docs.docker.com/docker-hub/usage/