the system hang

[Anh-Thu Tran]

My linux box was hanging twice this week. I can ping it, but can not telnet to it. Even on the login window, I could not type anything. I had to endup to power off and on. I could not find anything on the message log except it shows: Nov 17 09:19:39 fwcl7 kernel: nfs: too small RPC reply size (0 bytes)

and this is the last message before I restart the box. Any ideas why it happened like this? I am worry that it will happen again. Thanks.

[HGN]

Hi

We had a similar situtation more than a year ago,from the server and os perspective
1.Hardware : Ran diagnostics and everythign came good (downtime required)
2. OS : Contacted RedHat and they would only ask to go in for a newer kernel.

The problem was with the application which was tuned after many days and after that it has been working fine, the application on our sever was apache.

Rgds

HGN

[Anh-Thu Tran]

You mean you don't need to do anything and the box is doing fine now? We have apache (tarantella) running on it.

[Bill Thorsteinson]

Run top and look at the free memory.

It it is decreasing over time then eventually
the system will hang. UNIX depends on having
free memory in order to launch new commands,
logins, etc.

RPC reply size could be 0 because no memory
was available for allocation.

It looks like a memory leak of some sort.

You may want to tune apache to kill processs
after fewer requests. Default is about 1000,
but if you have code which is leaking memory
you may want to reduce this to 100.

Look for a memory leak.