BUG: "Test connection" always returns "Connections successful" about jenkins-plugin HOT 11 CLOSED

from jenkins-plugin.

Yep, I see this as well putting in a bad token.

Cross referencing with our extended test, the extended test for failure supplies a bad cluster api url to confirm test failure. We don't have a test that supplies a bad token in jenkins_pipeline.go.

from jenkins-plugin.

Perhaps the http get of healthz/ready does not require a valid token ?!?! I have verified we are passing in the configured (invalid) token.

I'll experiment with some other uri endpoints

from jenkins-plugin.

Yep, that appears to be the case. I added an equivalent of oc get is -n openshift and that produced an auth error.

Conversely, if I go with the default token (the serviceaccount's token), the oc get is -n openshift succeeds.

@bparees - does rest call equivalent to oc get is -n openshift seem to you like a valid rest call to determine whether the supplied token carries the necessary authorization ?

from jenkins-plugin.

is there no generic endpoint we can hit to simply validate the token? @liggitt @deads2k

from jenkins-plugin.

there is the "/oapi/v1/users/~" endpoint, but it is messier to access that one from the java restclient

from jenkins-plugin.

my worry w/ the openshift NS suggestion is the day someone decides they don't want an openshift namespace.

from jenkins-plugin.

ok, i managed to hit "/oapi/v1/users/~" with relatively little pain from the java restclient after all ... pending endpoint correction from @liggitt @deads2k let's go with that

i'll submit pull momentarily

from jenkins-plugin.

/apis will return 401 with a bad token, and exists in both openshift and kube servers

from jenkins-plugin.

@livelace if you can try out the pre-release version of this change at https://ci.openshift.redhat.com/jenkins/view/All/job/openshift-pipeline-plugin/36/s3/download/openshift-pipeline.hpi

from jenkins-plugin.

@gabemontero It works, thanks! :)

from jenkins-plugin.