fmsedge stops passing traffic on :80 (not :1935)

Adobe-

I'm a systems administrator at an ISP running many FMS 3.5.1 instances for an array of clients. About half of these FMS 3.5.1s fall over dead after 2-3 weeks of runtime.

Specifically, 'fmsedge' becomes unresponsive to http/rtmp tests on port 80; NOT on port 1935. The process appears to be doing work (cpu time is nonzero), and it answers the phone - I can connect to port 80. But I can't pass traffic to the daemon.

Logs give no indication anything's wrong. I can't strace or truss java with any legibility, due to fmsedge living atop a multithreaded virtualmachine.

This occurs regularly, but does not occur on /every machine/; only about half of them. Where it happens, it's at 2-3 weeks after process launch.

I am hoping you can advise me on how to collect additional diagnostics, so that I can find what's breaking in fmsedge and either prevent it, or ask you for a patch.

I am running on CentOS 5.3 (like RHEL-5), 64-bit; I install nothing but Adobe FMS, and used Adobe's installer. All my nodes are identical.

Regards and thanks in advance, I pray-

--

ecv@voxel.net