I have a similar issue with a pair of centrally managed 1550s running R80.20.50, operating in a ClusterXL cluster. The cluster is used for remote access for a small number of users via the Checkpoint mobile client. Every 3 to 11 days, all the clients start getting the exact same error. In my case, since the pair are centrally managed, pushing policy to them restarts the firewall daemon and resets the clock on the issue. So, we are proactively pushing policy to the cluster every three days to forestall angry users.
I had an open TAC case about the issue, which went on for several weeks before auto-closing without a resolution. The case's next steps would have me update the appliances to R81 to see if that fixes the issue, which I'm willing to do, but first have to arrange to upgrade our management system to R81, which is not quite as easy as upgrading the appliances. The fact that your appliance is already running R81 gives me some pause.
I am very much interested in what your outcome with TAC will be; please, please keep us posted on the findings. I realize our issues are not exactly alike, but hopefully there will be some cross-application. Hopefully I will get the time to upgrade our central management in the next week or two, and then i will reopen our TAC case.