I got bit more info...here is their response:
Failover is normally fine when initiated, however failback is usually a problem specifically when the link is utilized for extended period. If I failover straight away it seems to work ok.
I’m wondering if the tunnel expires during long periods and pings can’t make it across the tunnel when attempting to failback.
**************************************************************
I asked them for the network diagram that would make this even more simplified, but to me, sounds like when there is one fw with the issue or it has to be rebooted, failover to next one works fine, but then failback is where the issue occurs.
Andy