This is most likely a timing issue, thus it is usually not observable when using a local CDK but always reproducible on OpenShift Online:

steps:

1. ASSERT: have an application - with deployment configs, replication controllers, pods, etc.) running in OpenShift Online (ex. created via nodejs-mongo-persistent)
2. EXEC: launch "Delete Resources" wizard
3. ASSERT: you see deployment configs, replication controllers, pods etc.
4. EXEC: select all resources and hit "Finish"
5. ASSERT: wait for all removal jobs to be finished
6. EXEC: launch "Delete Resource" wizard again and inspect the resources that are still there

Result:

Replication controller(s) and pods are left in place, they were not deleted.
Some pods may be in terminating state and disappear eventually.