I've a big OpenShift Origin 1.1 cluster with many nodes. Now I've pulled an image from an insecure registry (self-signed certificates) on one of the nodes. I can perform the oc new-app command to start the image and the application. The problem appears when I try to scale.
The scaling itself is going fine but the new containers are all on the same node from where I've pulled the image.
I thought the problem was the following: I'm creating a new app from an existing docker image. So because there is no s2i-build it's not necessary to build a new image and more important: it's not necessary to push this image into the OpenShift registry.
So other nodes aren't able to pull the image when I try to scale because it's not in the registry. It's only locally on one of my nodes.
How is it possible to solve this problem? I tried to push my image into the openshift registry manually. This was possible but after scaling the container is still recreated on that same node where the image already is.
I was reading about image streams but I don't know if that will solve my problem. When I will try to create an image stream on my image, will the image stream be known over the whole cluster-environment?
Deployment configs point to images to run. If the nodes have access to that image already, you won't need to do anything except reference it in the DC pod template. If you want to control how the image is rolled out, you create an image stream and then have your DCs point to that image stream, then you can tag a new image in to the stream and rollouts will be triggered.