Glad to hear you got it fixed, one of our team managed to find the issue on our system today and it was down to routing as well but had to drop down to the systemshell to spot it (should de discoverable in the node shell as well). What was odd is that ping from the affected lif continued to work.
From the cluster shell the node management lif had the correct routes/costs/etc. assigned to it identical configuration to the other nodes (based on this I ignored that as being an issue) but a colleague took a closer look at the node in the systemshell and found there was no default route in the routing table, the output was compared to the other nodes to be sure and it was just this one node that it was missing on.
Resolution was just to delete the route in the routing group and recreate it and everything started working. I have requested that support create a KB on this as it is a bit obscure.