Deletion of Service/Binding CRD instances failed to delete service instance about cloud-operators HOT 10 CLOSED

in the service that remained, is there still a credential?

from cloud-operators.

yes. there was still a credential.

from cloud-operators.

what's weird is that the log actually indicates that the service was deleted without errors. The first time it fails, but it tries again and the second deletion went through without errors. The binding then is unable to delete because when the service deletion is successful the service and binding kube objects went away.

I am unable to reproduce. If there is any other detail you can convey, maybe the timing of operations. Are all the yamls submitted together? How long do you wait between creation and deletion?

from cloud-operators.

Are you sure that when you refresh the console in bluemix, the service is still there?

It is also possible that bluemix returned a 410 on the second deletion indicating that the service has been deleted previously. In that case, our controller would silently do nothing. In all other cases it would have returned and logged the error

from cloud-operators.

The service was still in bluemix 24 hours later. Multiple refreshes of the console page.

I created/deleted the resources relatively quickly. Service and Binding yamls were both submitted and deleted simultaneously (one kubectl command with combined yaml).

I noticed hours later (but still yesterday) that the service was still there in bluemix. I checked again today before opening the bug and the service was there. I have since manually deleted it (so I could recreate it again with the operator). So far today have created/deleted 3 times using the operator with no problems.

from cloud-operators.

ok thanks. I also asked the Watson Translator folks to understand under what situation we get the error The resource key cannot be deleted. Maybe that will give us a clue to reproduce.

from cloud-operators.

Had it happen again, this time when using an RHOS 3 cluster. Attaching log. I'll leave the translator service instance alive until Monday in case there is anything useful to get by examining it.
log22.txt.

from cloud-operators.

This was due to a bug in the way we were checking if a service doesn't exist upon deletion. The bug has been fixed in PR #41

from cloud-operators.

@dgrove-oss please let me know if we can close this issue now

from cloud-operators.

Have not seen problem since I updated to new version of operator. Closing.

from cloud-operators.