dns failing inside the container

[rncry]

We're using local consul agents to provide dns for all our nodes in the cluster. This is setup via resolv.conf and works without any problems both on the host and inside all other running containers.

mesos-consul however cannot seem to resolve hosts (namely zookeeper). I can ping zookeeper on the host and it resolves, I can start an ubuntu container and ping zookeeper from inside it. I've tried changing the entrypoint of mesos-consul to /bin/sh and pinging our zookeeper hosts fail inside the container. Any ideas?

[rncry]

It seems that search domains are not respected within the container. If I give the FQDN for zookeeper it works.

[rncry]

/ # ping zookeeper ping: bad address 'zookeeper' / # ping zookeeper.service.test PING zookeeper.service.test (172.20.121.251): 56 data bytes 64 bytes from 172.20.121.251: seq=0 ttl=63 time=0.535 ms ^C --- zookeeper.service.test ping statistics --- 1 packets transmitted, 1 packets received, 0% packet loss round-trip min/avg/max = 0.535/0.535/0.535 ms / # cat /etc/resolv.conf nameserver 172.17.42.1 search service.test node.test test

[rncry]

Apparently this is not supported in alpine linux's resolver: https://bugs.alpinelinux.org/issues/469

[flyinprogrammer]

call me crazy but it would seem the space savings of alpine is > than the pain of supplying a FQDN ?

[rncry]

In some of the environments we're building we can't provide a FQDN programatically so it's essential for us.

On Fri, Oct 30, 2015 at 1:22 AM, Alan notifications@github.com wrote:

call me crazy but it would seem the space savings of alpine is > than the pain of supplying a FQDN ?

— Reply to this email directly or view it on GitHub https://github.com/CiscoCloud/mesos-consul/issues/10#issuecomment-152373763 .

[flyinprogrammer]

fair enough!

[rncry]

Closing as there is now an ubuntu image provided :) thanks!

[erSitzt]

Switching the Dockerfile to Alpine 3.6 helps too

FROM golang:1.8-alpine3.6