Comment 7 for bug 1886429

If there are lingering armada pods in non-Running state (eg, Unknown, Terminating) then Dan's proposed solution should work. Ideally we should also cleanup stale armada pods, but that requires --force.

WORKAROUND: Forcibly delete all armada pods (note this is safe to do):

kubectl get pods -n armada | awk '{print $1}' | \
  xargs --no-run-if-empty -i{} kubectl delete pod -n armada {} --grace-period=0 --force