I'm trying to delete jobs using a filter like the following
kubectl delete jobs -l ml=""
This returns no resources found. However if I do kubectl describe I see
Labels: ml=,job_type=worker,runtime_id=tf-runtime,task_index=0
The same filter and command works just fine with the pods created by the job controller.
The command also works just fine my job is tagged with a single label, for example
Labels: ml=
So my filter appears to be incorrect when there are other labels on the resource. However, the same set of labels on other resources (services, pods) works just fine with that filter.
I just tried this with with kubectl 1.3.0 and I can't reproduce it. Can you try the latest kubectl?