Re: [tlug] strange nfs crashes

[Michal Hajek <hajek1@example.com>]

Hello, 

after some time, I am resurrect this thread. So far we used some
workaround our problem with NFS and new switch. However, yesterday I
have finally made the suggested test:

* Kalin KOZHUHAROV (me.kalin@example.com) [100415 01:57]:
> tcpdump -i <interface> -s 65535 -w <some-file>

I did it on server and client machine as well:
tcpdump -i eth1 -w /tmp/file.tcpdump

(-s 65535 is the default) 

Unfortunately, before the application (the one I am using for
calculation, _not_ tcpdump itself) crashed, I got two roughly 10GB files
- one for client, one for server.

Now there is my problem: how do I analyse it? 

wireshark causes the machine to start swapping and become unusable. This
happens even on 8GB RAM machine.  Suggested pcapdiff [1] behaves about
the same way.

Any ideas?


[1] http://www.eff.org/testyourisp/pcapdiff