bge watchdog timeout - resetting

[Eugene M. Zheganin]

Hi,

I'm having troubles with one FreeBSD server, running 8.2-STABLE.
Randomly I get 'bge watchdog timeout - resetting' errors on it's 
console. It can run 1-2 months without problem, then I can get these 
errors like twice per day. They are appearing in bunches, and the system 
becomes irresponsive: I can see it's alive on the console, I can input 
the text, but I cannot login. Then I have to reset it. The system is 
also partially dropping the traffic (I mean partially receiveing and 
processing) - I have an identical box with carp running, but it cannot 
become MASTER, as it seems like some part of the traffic is handled by 
the troubled device. This box is an IBM x3250m2 server, I have like two 
dozens of this, and this is like the most distrurbing one. I'm also 
getting these messages on another box, also running 8.2-STABLE, but 
others are doing just fine (they are running mostly 8.2-RELEASE or even 
7.x FreeBSDs, but I also have a couple of STABLEs running just fine).

What can be done here  (can something be done) ?
Is it worth to change the cable or the catalyst port (did it help 
someone ?) ?

Would it help to turn on the hw.bge.allow_asf ?

the bge0 is (as the pciconf reports is):
Broadcom NetXtreme BCM5722 Gigabit (94309)

I can also say that it's running one of the recent firmwares, as I 
updated it already from IBM bomc, in order to resolve the situation, but 
didn't help much.

Thanks.
Eugene.