Now, you're lucky.
I've had similar problems with a 600C running 5.0.9 for more than a year. After around 100 days port status 'disappeared' from the 'Unit Operation' widget, traffic history widgets were showing the exact same levels on all ports (which couldn't be true). CPU load was 100 % (in the GUI only!). Apart from these cosmetics no functions were compromised. A reboot "solved" the problem for another 60 to 100 days.
Without reboot the 600C would eventually freeze completely - no way to reboot except for a power-cycle, with partial loss of the configuration. Not good.
Other fellow forum members were experiencing the same: https://forum.fortinet.com/tm.aspx?m=122740
It turned out that there is an incompatibility of the BIOS or firmware with the memory modules on a 600C. Only the 600C is affected. A BIOS upgrade will not help much (it will only allow for a warm reboot instead of a cold one) but a special build of v5.0.12 will.
Finally, we installed special build 8408 and haven't had any problems since (uptime: 173 days). The bugIDs involved were 0229553 (NP4 failing to get the switch) and 218033 (new switch SDK), and apparently, 0243461 as well (see post mentioned above).
Finally, in v5.2 Fortinet has incorporated the fix:
in v5.2.4: "for FG-600C, PC: 08908-08 and later, memory incompatibility, fixed in v5.2.2 and later"
but there's no direct reference to the 'switch' problem.
Our FG-600C is PN 08908-02, so older than that, so it's probably not affected by the memory bug but the switch bug.
I'd open a Support call with FTNT and refer them to our case (sorry, no caseID av. at the moment), or the special build of v5.0.12. That should solve your problem within half a day.
Ede Kernel panic: Aiee, killing interrupt handler!
