Description of problem:
Default etcd image used with containerized installations is registry.access.redhat.com/rhel7/etcd instead of rhel7/etcd which forces using osm_etcd_image=rhel7/etcd with disconnected installations.
Please consider dropping the hard-coded registry name from the default as is the case with other images already.
Version-Release number of selected component (if applicable):
If we do that for Centos and Fedora it will pull docker.io/rhel7/etcd which doesn't exist. If we did the same substitution on registry.access.redhat.com/rhel7/etcd that we do for image streams and templates will that work for you? It's certainly getting confusing either way.
Good point re CentOS/Fedora - yes I think the substitution might be the most suitable option here, at least based on my current understanding. Thanks.
Scott, can we get this change implemented we are more and more running into it specially after we focus on image based installs instead of rpm based installs.
and when osm_etcd_image is not listed in the hosts.examples it gets hard to find. Of cause now we know, but lots of people out there dont..there is a PR for the hosts.example file so it will display osm_etcd_image, but a more generic change for oreg should be made.
Ture have mad a pr https://github.com/openshift/openshift-ansible/pull/6618
Merged the examples documentation PR, we can work on this for 3.9.
*** Bug 1505646 has been marked as a duplicate of this bug. ***
Checking the example hosts file this seems to be reflected in 3.7+.