web console does not recognized failed provision

[bparees]

If i do a provision operation that fails (jenkins-ephemeral currently fails due to permissions issues), the SC instance object reports:

  status:
    asyncOpInProgress: false
    conditions:
    - lastTransitionTime: 2017-06-14T05:11:30Z
      message: 'Provision call failed. Error provisioning Instance "p3/jenkins-e
phemeral-fln2k"
        of ServiceClass "jenkins-ephemeral" at Broker "template-broker": Service
 instance
        with same id but different attributes exists'
      reason: ProvisionCallFailed
      status: "False"
      type: Ready
    lastOperation: provisioning

and the templateinstance reports:

  status:
    conditions:
    - lastTransitionTime: 2017-06-14T05:10:54Z
      message: User "system:serviceaccount:openshift-infra:template-instance-con
troller"
        cannot create rolebindings in project "p3"
      reason: Failed
      status: "True"
      type: InstantiateFailure

but the web console just spins saying "provisioning", it doesn't seem to recognize the provision has failed already.

[bparees]

(it's possible the SC needs to be making the state of the instance clearer, so this might need to go to @pmorie instead)

[jwforres]

We have the same question for @pmorie that we had for binding in openshift/origin-web-console#1466 (comment)

Is this a terminal condition for the provision. Can we rely on this condition meaning that provision is perm failed.

[jwforres]

cc @spadgett

[bparees]

Sounds like there are two issues for @pmorie here:

1. the SC should not be retrying the provision since the broker told it it failed.
2. the SC is not marking the provision to a failed state that the web console can understand.

[spadgett]

We could show an error for reason: Failed if it's the right thing to do. Right now we're just looking at type: Ready / status: False and showing the pending message.

I think we should show the ready condition message somewhere in the UI if available since it's likely helpful, maybe in the expanded provision service row on the overview.

[jwforres]

agreed but we still need to know without question what the permanently failed reason code is for both provision and bind

…

On Wed, Jun 14, 2017 at 2:05 PM, Sam Padgett ***@***.***> wrote: We could show an error for reason: Failed if it's the right thing to do. Right now we're just looking at type: Ready / status: False and showing the pending message. I think we should show the ready condition message *somewhere* in the UI if available since it's likely helpful, maybe in the expanded provision service row on the overview. — You are receiving this because you were assigned. Reply to this email directly, view it on GitHub <#14629 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/ABZk7asq-FP3UMWSxurT0QXx2ZdILXzwks5sECD4gaJpZM4N5ZG5> .

[jwforres]

relevant changes in service catalog API https://github.com/kubernetes-incubator/service-catalog/pull/1017/files

[spadgett]

Fixed by

openshift/origin-web-catalog#393
openshift/origin-web-console#1961