Docker-nginx: Alpine base cannot resolve links via DNS

[rocketraman]

Apparently there is an issue with DNS resolution in alpine. See: https://forums.docker.com/t/container-link-problem-in-docker-cloud/6445/12. This prevents nginx images from resolving linked containers via DNS, such as on Docker Cloud.

Using alpine:edge will apparently fix the issue.

[thresheek]

I volunteer @tianon to decide if it's worth doing now.

[tianon]

Heh :innocent:

IMO, since Alpine upstream doesn't have the fix in a release yet, we're
probably better off waiting for the next major release of Alpine since this
will be fixed for us then ("edge" is essentially Alpine's "unstable" or
"devel" suite)

[rocketraman]

That makes sense. FYI, I can confirm rebuilding the container with alpine:edge as a base does indeed fix the issue, making container links work correctly on Docker Cloud. Hopefully alpine will release soon!

[Naddiseo]

Is this fixed with the recently released alpine 3.4?

[Krijger]

This is not fixed in 3.4, no.

[thresheek]

The nginx:alpine image was rebuilt with Alpine 3.4. Please test.

[genxstylez]

not fixed

[hensansi]

For me it's fixed using 1.11.4-alpine,
Nginx is able to resolve upstreams and I am able to resolve names using host.

[thresheek]

Can anyone confirm if it's really fixed? Thank you.

[andreif]

1.11.13-alpine not fixed.

[thresheek]

Please test the nginx:alpine which is now based off alpine:3.5 to see if it's still an issue.

[andreif]

Works fine now for 1.13.0-alpine on macOS.

[leialexisjiang]

I just tried 1.13.0-alpine and 1.13.3-alpine, it seems the problem is still there.

[tnguyen14]

I'm seeing this problem with 1.13 alpine as well, not RHEL 7.3

[hirokishirai]

Are there any updates?

[leialexisjiang]

just tried with docker-nginx version debian which works fine, but version alpine still does not work, it seems that version alpine uses still alpine 3.5 which has DNS problem.

[andreif]

FROM alpine:3.5

https://github.com/nginxinc/docker-nginx/blob/master/mainline/alpine/Dockerfile#L1

I wonder how it worked for my case in April though. I am not working with it right now so won't be able to test it again.

[vedmant]

It looks like problem is still here, so we can't use apline for swarm?

[J0WI]

It works for me, but only when I add resolver 127.0.0.11; to the nginx config.

[boatcoder]

I'm going to guess that the reason people are seeing different results is because of this tiny bit of config and the environment in which they are working....

  resolver 127.0.0.11 valid=30s ipv6=off;
  resolver_timeout 5s;

So if your environment doesn't have ipv6 available, the resolver will fail almost instantaneously and you will lose some of your hair trying to figure out why.....

[thresheek]

A way to test:

$ docker run -n app nginx:1.19.7-alpine
ceaa16117ed4b767ce1b045effaca14e154d002443e43ae4745e51af28396d0a
$ docker run --name web --link app:app nginx:1.19.7-alpine ping -c1 app
PING app (172.17.0.2): 56 data bytes
64 bytes from 172.17.0.2: seq=0 ttl=64 time=0.150 ms

--- app ping statistics ---
1 packets transmitted, 1 packets received, 0% packet loss
round-trip min/avg/max = 0.150/0.150/0.150 ms

So it works in current images. Closing it.