Red Hat Bugzilla – Bug 1285536
cannot rsh using cluster admin on master node
Last modified: 2016-01-08 12:41:59 EST
Description of problem: Not sure I've got component correct. I observe a strange thing. Given a 2 node cluster, I create a pod as a regular user. Then ssh to master and try to rsh it. That fails. While running the rsh command on a remote machine as the regular user does work just fine. Version-Release number of selected component (if applicable): oc v3.1.0.4-9-g72d3991 kubernetes v1.1.0-origin-1107-g4c8e6f4 How reproducible: always Actual results: # oc rsh cakephp-example-1-cutxg --namespace=c48hh -vvvvv Error from server: pods "cakephp-example-1-cutxg" not found Expected results: the bash prompt
Jordan please take a look here, does this look like authorization?
I think the namespace argument is not being honored. If I recall correctly, args to rsh need to come before the pod name. Can you rerun with --loglevel=6 to confirm which namespace is being used? You might need to do this: oc rsh --namespace=c48hh cakephp-example-1-cutxg
I'm almost certain that is the issue, since -vvvvv isn't valid. You'd need to put --loglevel=… immediately after the "rsh" as well
my bad, didn't realize rsh also can execute command given on command line like `exec`. In any case would be good to clarify command synopsis: > oc rsh POD [command] [options] One can assume `options` is for `oc` options. `exec` command help is a little more clear although it would be good to tell users where to put `oc` options. Should we leave this bug open for documentation or close as notabug?