connection problem

[hanyyu1]

There are about 20 unix servers in my network , but sometimes the servers may be disconnect due to the unstable network or some other unexpected reason , this make some problem to our work , so I want to know once the connection has problem and know what server ever has connection problem ( eg. at night time ) , could suggest a way ( or script ) that can do it ? thx.

[Torsten.]

First of all you need to find the reason for your unstable network and resolve the issue (bad cables, bad switches, duplicated IP addresses ...).
Have a look at the console output and syslog.log of your servers and search for network related messages like "network disconnected due to ...".

[RAC_1]

The problems about network are logged at different files.

To start with you can look at nettl.LOGxx files in /var/adm
netfmt -f /var/adm/nettl.LOGxx
Also syslog.log, dmesg

glance metrics about network would also be helpful.

[saju_2]

Hi

Is there any hp-ux, linux,solaris machine which is unaffected with this network trouble? if so, put cron job for pinging all theses 20 machines in every minute. Redirect the output to some file. Do this for one day and observe the results..

Check the syslog for any kind of network related errors.

Network information regarding ports, ip's and cables should always be documeted and maintained properly, else it will be the biggest nightmare for any administrator.

Regardsd
CS

[hanyyu1]

thx replies ,

I will check the reason of disconnect , but is there a way such as ping from one of these servers to each servers , if found any server is diconnected , then inform me by mail , is it possible ? thx.

[Mark Ellzey]

Hanyyul,

Here's a script that will check a list of servers and 1) see if they are up, and 2) see if any filesystem is over 90% full. You can modify it for your particular situation.

I have to agree with the previous answers, though; you need to find out why your network is so flaky. Systems shouldn't just disconnect for no reason.

Regards,
Mark

[Mark Ellzey]

P.S.

You may also want to change the remsh commands to ssh, for security reasons.

Mark

[AwadheshPandey]

see this script and modify it as per ur needs,

Awadhesh