On Wed, May 18, 2016 at 10:14 AM, James Falkner <[email protected]> wrote:

> Is there any reason (other than bugs :) ) that pods instantiated to
> execute builds of Docker-based apps wouldn't be able to resolve
> *.svc.cluster.local names? I have two apps in a project, one built using a
> Docker strategy, and one using s2i.. the s2i one is able (at buildtime) to
> contact other services in the cluster using foo.bar.svc.cluster.local but
> the Docker one cannot (it is trying to use a local Maven mirror and failing
> to resolve the nexus server hostname using the *.svc.cluster.local name).
> Using the externally-exposed name works fine.
>

​i'm actually surprised either of them works.  Both the assemble script,
and your dockerfile commands, are running inside a container that's been
launched directly on your node host, meaning the /etc/resolv.conf that's
available to the container is the host's /etc/resolv.conf and it doesn't
have the openshift skydns ip injected.  The assemble container may have DNS
access because it picks up the networking config from the actual build pod,
which I guess may include DNS config, not sure.

the accepted workaround currently is to add the skydns ip to your host's
/etc/resolv.conf.

​


>
> Thanks!
>
> -James
>
> _______________________________________________
> users mailing list
> [email protected]
> http://lists.openshift.redhat.com/openshiftmm/listinfo/users
>



-- 
Ben Parees | OpenShift
_______________________________________________
users mailing list
[email protected]
http://lists.openshift.redhat.com/openshiftmm/listinfo/users

Reply via email to