[X-Unix] How to determine which network component is failing

[Charles Howse]

On Jul 24, 2006, at 12:27 PM, Wing Wong wrote:

> On 7/23/06, Charles Howse <chowse at charter.net> wrote:
>> Hardware setup:
>> ISP: Charter Communications, 2 megabit service
>> Cable Modem: Motorola SB4200 Surfboard
>> Router: Linksys Etherfast(r) Cable/DSL Router BEFSR41 V3 Firmware
>> Version: 1.04.17
>> Computers: 2 FreeBSD 6.1-RELEASE. 1 Mac OS X 10.4.7, 100 baseT
>> ethernet connections on all
>>
>> This morning, I was offline, and just by accident decided to try and
>> release and renew the ip address using the web interface of the  
>> router.
>> BINGO!  Online again!
>>
>> Is this an indication that the router is not noticing that the lease
>> has expired on my dhcp assigned ip address from the modem?  Or is it
>> an indication that the modem is not renewing it's ip address?  Or is
>> it something else?
>>
>> Can I tell the router to release / renew the ip address from a shell
>> script?
>
> Yes. I have a WRT54G Linksys router hooked up to my cablemodem and
> this does indeed happen once in a while. Basically, when this happens,
> one of two things happen:
>
> 1) the cablemodem/service has died and the cablemodem needs to be  
> power-cycled.
> 2) the dhcp lease needs to be renewed.
>
> Typically, I've found the lease to work fine by itself. But from time
> to time, it doesn't. :/ The reasons for this vary, but I'm guessing
> it's someone bouncing something over at the cable provider side and
> the lease is just "forgotten", leaving your router with a useless
> lease. So, hitting the release/renew will let you grab a new useful
> lease, and back on the net.
>
> I suppose you could script a lynx application to login to the web
> interface of your Linksys router and perform a lease release/renew
> once in while or when it is unable to access some remote server on the
> internet.
>
> Ie,
>
> 1) am I able to see the world?
> ** check google.com
> ** check msn.com
> ** check yourfavoritesite.com
> If at least one is up, then you are essentially live and the other two
> sites are coincidentally dead.
>
> 2) If I'm not up, let's run the scripted lynx session to perform a
> release/renew of my IP address on the router.
>
> 3) Check output of the script. Did I actually get an IP address or am
> I still stuck with 0.0.0.0? Wait a bit longer and run the scripted
> lynx to read the DHCP address from the router again. If fail yet
> again, scream for help!!
>
> If you have some means to, you could script a powercycle of the cable
> modem itself with an intelligent power socket. (serial control or over
> net control) Just make sure you don't powercycle more frequently than
> necessary. I've blown devices that way. ;) If you powercycle every
> 30seconds to every 5 minutes, it's amazing what 2 solid weeks of power
> cycling will do to a device.
>
Here's what I have so far (not working to suit me), but we haven't  
thought out all the scenarios yet.  Not ready to release/renew the  
dhcp lease nor reboot the modem yet.

Example: I ran what I have so far with the modem in standby, and  
terminal returned:
[charles at larry:~]$ bin/online.sh
ping: cannot resolve google.com: Unknown host
That makes sense, but the $logfile now contains:
Tue Jul 25 18:19:30 CDT 2006
host1 online, exiting
That's because the script didn't return any ping results, just the  
failure to resolve the name.
I need help with:
1) what if my 3 pings returns only 1, or 2?  That equals online, but  
it's easier to test against $offline (below).  Therefore, the script  
will think the host is offline unless it returns all 3 pings.




#!/bin/bash

logfile=Documents/online.txt
stats=`curl -s http://192.168.100.1/signaldata.html | grep dB | tr -d  
\<TD\> | tr -d /`
offline='3 packets transmitted, 0 packets received, 100% packet loss'

what_failed () {
	echo >> $logfile
	echo OFFLINE >> $logfile
       	date >> $logfile
	ping -c 3 localhost | tail -3 >> $logfile
	echo >> $logfile
	ping -c 3 larry | tail -3 >> $logfile
	echo >> $logfile
	ping -c 3 moe | tail -3 >> $logfile
	echo >> $logfile
	ping -c 3 router | tail -3 >> $logfile
	echo >> $logfile
	ping -c 3 modem | tail -3 >> $logfile
	echo >> $logfile
	echo "$stats" >> $logfile
}

if [ ! "`ping -c 3 google.com | grep packets`" == "$offline" ] ; then
	date >> $logfile
	echo host1 online, exiting >> $logfile
	exit 0
elif
	[ ! "`ping -c 3 microsoft.com | grep packets`" == "$offline" ] ; then
	date >> $logfile
	echo host2 online, exiting >> $logfile
	exit 0
elif
	[ ! "`ping -c 3 yahoo.com | grep packets`" == "$offline" ] ; then
	date >> $logfile
	echo host3 online, exiting >> $logfile
	exit 0
else
	what_failed
fi

exit 0

--
Thanks,
Charles
http://bubbabbq.homeunix.net