ningu in #wikimedia-cloud reported that webservice --backend=kubernetes node10 shell was failing with "Pod is not ready in time". I asked them to try it again, and it timed out another two times. I logged in to Toolforge and tried to open a node10 shell, a nodejs shell, and a python3.7 shell, all of which failed the same way. I tried the node10 shell and the python3.7 shell from a tool I have migrated to the new k8s cluster (anticompositetools) and one I have not (anticompositetest), and it failed in both cases.
webservice --backend=kubernetes [container] {start|stop|restart|status} is unaffected and works normally.