Orphaned mitigation on service instance update failures

brunograz commented 4 years ago

Thanks for submitting an issue to cloud_controller_ng. We are always trying to improve! To help us, please fill out the following template.

Issue

CC issues an orphaned mitigation on service instances with failed updates resulting in the removal of the service instance.

Context

Updating service instances may result in a failure and for such scenario the orphaned mitigation should not be triggered as described in the OSB specification.

We assume that this commit has introduced this bug where it triggers the mitigation when a request to last_operation is issued by the CC and the service broker report its state is 'failed'.

https://github.com/cloudfoundry/cloud_controller_ng/commit/80176ff0068741088e19629516c0285b4cf57ef3#diff-6e29198ceec350cf6dfa0220146b0c61

From our point view it is incorrect to initiate a mitigation on a service instance based on the last_operation without checking the type of the operation that had been performed.

For additional information:

The service instance provisioning / deprovisioning / update is asynchronous.
The service broker returns a http 200, state = 'failed' when last_operation is requested by CC.

Steps to Reproduce

Create any service in Cloud Foundry
Update the service (change of a plan, configuration)
- The service update must fail

Expected result

Update service instance fails but no orphaned mitigation is triggered.

Current result

Update service instance fails and orphaned mitigation is triggered.

cf-gitbot commented 4 years ago

We have created an issue in Pivotal Tracker to manage this:

https://www.pivotaltracker.com/story/show/173291140

The labels on this github issue will be updated when the story is started.

blgm commented 4 years ago

Thank you for raising this @brunograz. This is clearly an incorrect behavior and we are working on a fix.

cloudfoundry / cloud_controller_ng