[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: Snapshot for debugging network problems



Hmmm, 'ipchains -L' or 'ipchains -L -v' could be useful
here. If the i/o chains filter or accept packets that
could be (part of) the problem you're a step closer to
the solution.

I use ipchains sometimes to block/accept packets to see
what will happen in some circumstances.

I'm very curious what will come of this thread...

Regards,

Onno

At 02:48 PM 11/5/99 +0000, David Wright wrote:
While tracking down network problems of any kind, it's quite handy
to take a snapshot of the networking parameters so you can look at
it after the event. I have a bash function which is currently:

    /bin/uname -a
    /sbin/ifconfig
    /sbin/route -n
    /usr/sbin/arp -n -a
    /bin/netstat -n -a -e
    /bin/ps auxwww
    /bin/date

Then I use
    tcpdump -l -n -i <interface> [host <host>] | tee <somefile>
to watch the traffic and
    /bin/fuser -n <udp or tcp> -v <port number>
to see what might be causing trouble. The last one I really stumbled
across, only having seen fuser used for investigating busy files and
directories in the past.

Are there any useful commands I've missed? What's the best tool for
translating the output from tcpdump?        [stop here if you like]

For those that might be interested, the last problem I was trying
to solve was a dramatic slowdown in ppp from my home machine to work.
So slow that ssh just wouldn't connect, and telnet would take more
than five minutes. Characters could take up to a minute to reflect.

I was at work and had initiated the ppp connection. Looking at the
traffic with tcpdump -l -i ppp0, it was completely dominated by traffic
to the nameservers, and with the fuser command on the port numbers
being shown, I was able to pin it down to icmplogd and tcplogd which
were running on the m/c at home. (No longer.)

I have no idea why this slowdown had happened only a couple of times
in the past, but previously I'd put it down to a bad line or some
ethernet problem at work. (It never seemed to affect the CHAP
handshaking, though.) I attacked the problem this time because I was
sitting at the work end, so I could easily confirm that everything
on the ethernet was functioning. (And I really needed to transfer a
file home.)

It took me most of an hour to realise I should kill the two offending
daemons. I'm still not sure what they were asking the nameservers,
but I have a large traffic file available. I'm used to seeing messages
like (from memory) "bar > 255, who is foo, tell bar" and "foo > bar,
foo is on 0:1:2:3:4:5" but this stuff was all numerical. Is there
something that can print what it thinks it all means?

Cheers,

--
Email:  d.wright@open.ac.uk   Tel: +44 1908 653 739  Fax: +44 1908 655 151
Snail:  David Wright, Earth Science Dept., Milton Keynes, England, MK7 6AA
Disclaimer:   These addresses are only for reaching me, and do not signify
official stationery. Views expressed here are either my own or plagiarised.


--
Unsubscribe? mail -s unsubscribe debian-user-request@lists.debian.org < /dev/null


Reply to: