ppc64le上的openshift原点

Question

I am trying to run openshift on ubuntu 16.04 ppc64le (IBM POWER8).

I succeeded in building the binaries from source (openshift origin), and in starting openshift.

However, when I did "oc cluster up", I see it pulls openshift/origin:v3.6.1 image from internet and it eventually fails.

I think Redhat pushed openshift/origin:v3.6.1 for x86 architecture only, not for ppc64le.

How can I configure it to use openshift/origin-ppc64le:v3.6.1, provided I build the image on my own ?

Below is the error I get with "oc cluster up".

u0017649@sys-90043:~/origin/_output/local/bin/linux/ppc64le$ sudo ./oc cluster up
Starting OpenShift using openshift/origin:v3.6.1 ...
Pulling image openshift/origin:v3.6.1
Pulled 1/4 layers, 26% complete
Pulled 1/4 layers, 51% complete
Pulled 2/4 layers, 73% complete
Pulled 2/4 layers, 82% complete
Pulled 2/4 layers, 89% complete
Pulled 3/4 layers, 97% complete
Pulled 4/4 layers, 100% complete
Extracting
Image pull complete
-- Checking OpenShift client ... OK
-- Checking Docker client ... OK
-- Checking Docker version ... OK
-- Checking for existing OpenShift container ... OK
-- Checking for openshift/origin:v3.6.1 image ...
   Pulling image openshift/origin:v3.6.1
   Pulled 1/4 layers, 26% complete
   Pulled 1/4 layers, 51% complete
   Pulled 2/4 layers, 73% complete
   Pulled 2/4 layers, 82% complete
   Pulled 2/4 layers, 89% complete
   Pulled 3/4 layers, 97% complete
   Pulled 4/4 layers, 100% complete
   Extracting
   Image pull complete
-- Checking Docker daemon configuration ... OK
-- Checking for available ports ... FAIL
   Error: Cannot get TCP port information from Kubernetes host
   Caused By:
     Error: Docker run error rc=1
     Details:
       Image: openshift/origin:v3.6.1
       Entrypoint: [/bin/bash]
       Command: [-c cat /proc/net/tcp && ( [ -e /proc/net/tcp6 ] && cat /proc/net/tcp6 || true)]
       Error Output:
         standard_init_linux.go:187: exec user process caused "exec format error"

Answer 1

If you built the binaries then you should be able to use the hack/build-local-images.py script to shove those binaries into local images. When you run oc specify the version and it should start with your images you just created: oc cluster up --version latest .

This works for me on x86, YMMV

Answer 2

Rather than label your built image openshift/origin-ppc64le:v3.6.1 you could label your built image openshift/origin:v3.6.1 tricking the sudo ./oc cluster up to your ppc64le image.

Attempts seem to have been made previously to get a multiarch image going https://github.com/openshift/origin/issues/17468