Ping Check Failing

[Wojtak, Greg]

I had set up a ping check on a server to kill off a specific process
if the network becomes unavailable, in a nutshell, because the app
is clustered and prone to split-brain.

The check works perfectly and the kill works exactly the way we wanted
it to and in our tests we were able to have our app fail over correctly
100% of the time.

However, I started seeing notices in my monit.log:

error    : ICMP echo response for 10.101.14.1 1/10 failed -- received 40
bytes, expected at least 48 bytes

Does anyone know enough about the underlying ICMP protocol to tell me why
monit is seeing a truncated response?  I have monit configured to kill
off the app after 10 ping failures, and this is counting against that
total, and needlessly killing off the processes (a running ping at the
command line show 0 dropped packets, no network errors in netstat, etc).

I'm running monit 5.3.2 on RHEL 5 for this particular configuration.

Thanks!


-- 
Greg Wojtak
Senior Unix Systems Engineer
Office: (313) 373-4306
Mobile: (734) 718-8472