Cat 2950's running out of memory

[cscott]

I have a cluster of Cat 2950's running 12.1.19 EA1 that are running out of memory after 2 or 3 days running. It seems they are running out of processor memory first, then I/O memory. A show mem dead shows a LOT of dead processes still in memory. I'm relatively new to troubleshooting LAN switching problems, so is this normal? The show mem dead command returns about a 7 page output of dead processes..with the following numbers at the bottom (which I dont know what they represent, im guessing the amount of memory taken by these dead processes)

------

22260

======

308340

Is anyone else experiencing this problem? It seems it didnt start until we upgraded to 12.1.19 EA1.

Thanks in advance.

[nikhil_m]

I don't really know if you should Zero down directly onto the dead Processes. I feel the problem is not this....this could be just a side effect.

[cscott]

I started tracking the active processes memory usage over a few days and figured out that the CDP Process would not stop growing. I opened a case with TAC and sent them my 3 days worth of sh mem summary outputs and they said this was a known bug on the 2950G's but now also apparently effects the 2950T's as well. They said to either take the switches out of the cluster and disable CDP or to downgrade the IOS. I took them out of the cluster and disabled CDP and rebooted the switches and the problem is now fixed.

[subbarao.s]

hi,

have look on below bug id...

CSCec89172