Conectivity Issues: service_died * fcrsgx * 12547 in listener.log

Question

This is a 10g database (10.2) monitored through Grid Control. Every
once in a while we get a TNS timeout on the EM side, and at the same
time we get something like this in the listener.log:
02-JAN-2009 21:04:45 * service_update * fcrsgx * 0
02-JAN-2009 21:08:30 * service_update * fcrsgx * 0
02-JAN-2009 21:08:30 * service_died * fcrsgx * 12547
TNS-12547: TNS:lost contact
02-JAN-2009 21:08:42 * service_register * fcrsgx * 0
02-JAN-2009 21:08:54 * service_update * fcrsgx * 0
Now, this outage is noted by CT.
We are opening an SR with Oracle Support, but they work VERY slow, so
perhaps someone here has an insight into this problem.
Does anyone know what the cause could be?
TIA

Answer 1

01 - Is there some kind of a IP Scanner/Firewall running, which keeps scanning port/address of the listener? For this you need to check for the TCP timeout parameter in your firewall.
02 - If not, then you need to contact Microsoft for a solution to the "Slammer Worm", as a result of which database traffic thrown on port "1434" is thrown away and dropped off.
03 - Have you verified if the "listener.ora" file has reached or crossed its maximum size limit? If this is the case then you need to simply archive or clear/delete the file and restart the listener.
04 - Lastly, check if your machine has more network card, which are on the same subnet. Switch off the one that isn't in use.
Hope this helps.