[c-nsp] Something like MTR, but forced path

Tuc at T-B-O-H.NET ml at t-b-o-h.net
Wed Dec 19 13:39:55 EST 2007


Hi,

	This was one of the ways I found out that this was happening. But I
want something I can see happening THE SECOND it occurs. A can SEE a ping not
reach a destination and come back. I have to WAIT for the system to go through
its paces and tell me this is happening.

	I'm looking for something UNIX based so it can be deployed in
various places so I can prove I'm not the only one seeing it. 

	Every time the event happens we grab more and more info, and
because his graphing doesn't always catch it, and even though he swears
it isn't gospel but his only point of defense is the graphs, I need
something that is real time. So far I've found that everything behind a
specific wireless backhaul is still up, infact I can see other client
routers. Luckily, one of them is DD-WRT, so I'm able to run fping (BEST
I got for now) and he is warming up to it being a way to believe what
I keep telling him (Over and over and over) is true. So now I have
it on :

FreeBSD himinbjorg.tucs-beachin-obx-house.com 5.5-STABLE FreeBSD 5.5-STABLE #9: Tue Jan 23 23:44:47 EST 2007     root at himinbjorg.tucs-beachin-obx-house.com:/usr/obj/usr/src/sys/HIMINBJORG53  i386

	and

Linux landolfrtr 2.4.34-pre2 #170 Fri Sep 15 20:10:21 CEST 2006 mips unknown

	if it shows at landolfrtr also, I think he might start to
believe me. The next thing is some way that when it happens at 
his end he'll believe it and be able to do a test he believes.
We were on the phone once when it happened, he told me that
while I was telling him I couldn't see the world, he could see
me , knew what the RSSI was, how much bandwidth was going, etc. 

	I don't know if he'll let me set up the monitoring on
his router and believe it, but its worth a try.

		Thanks, Tuc
> 
> You can't do this by using pooling or interval based monitoring system. You
> need to work on syslog or event based traps. I believe that your equipment
> at both end is Cisco. 
> 
> You must track reachable IPs and generate syslog or event traps if the next
> hope or whatever IP is unreachable. Here is an example
> 
> Here what you want to monitor: 
> 
> ip sla monitor 1
>  type echo protocol ipIcmpEcho 192.168.75.3 source-ipaddr 192.168.3.254
>  timeout 2000
>  threshold 2500
>  frequency 3
> 
> Here I'm defining time domain:
> 
> ip sla monitor schedule 1 life forever start-time now
> 
> 
> Here I'm starting track:
> 
> track 123 rtr 1 reachability
> 
> 
> You will find logs as show below on every event.
> 
> *Dec 19 12:53:19.204 PKT: %TRACKING-5-STATE: 123 rtr 1 reachability Up->Down
> *Dec 19 12:53:24.204 PKT: %TRACKING-5-STATE: 123 rtr 1 reachability Down->Up
> 
> For same you can also use Cisco Embedded Event Manager 
> 
> http://www.cisco.com/en/US/products/ps6815/products_ios_protocol_group_home.
> html
> 
> Regards,
> Masood Ahmad Shah
> BLOG: http://www.weblogs.com.pk/jahil/
> 
> 
> 
> -----Original Message-----
> From: cisco-nsp-bounces at puck.nether.net
> [mailto:cisco-nsp-bounces at puck.nether.net] On Behalf Of Tuc at T-B-O-H.NET
> Sent: Wednesday, December 19, 2007 6:38 PM
> To: Ed Ravin
> Cc: cisco-nsp at puck.nether.net
> Subject: Re: [c-nsp] Something like MTR, but forced path
> 
> > 
> > On Tue, Dec 18, 2007 at 09:01:50PM -0500, Tuc at T-B-O-H.NET wrote:
> > > I'm basically looking for something I can run on Unix and
> > > give me a curses view of IPs I give it to ping at the same time.
> > 
> > You could use Mon:
> > 
> >   http://mon.wiki.kernel.org/index.php/Main_Page
> >
> 	(Info deleted for brevity) 
> > 
> > > Curses MTR bolds the lettering when it has a ping loss which I want
> > > to catch my eye.
> > 
> > In my shop, we have Mon set to write to our terminals when it has
> > something important to tell us.
> > 
> > > As an FYI, the problem I have is I'm having connection losses between my
> > > site and the wireless WISP's gateway. I think the packet is getting to
> the
> > > backhaul link at the site here, but not to the other end of the backhaul
> here.
> > > I want to run a set of pings [...]
> > 
> > That sounds like a job for Smokeping:
> > 
> >   http://oss.oetiker.ch/smokeping/
> >
> 	(Info deleted for brevity)
> > 
> > > I want something formalized since sending pings that just show
> > > a loss somewhere in the middle don't mean anything.
> > 
> > I've done that with Smokeping - tell Smokeping to ping everything in
> > a particular path.  Lining up the resulting graphs with records of
> > service outages is usually very informative.
> > 
> 	I appreciate your reply.
> 
> 	I actually have smokeping running on both ends, but I'm looking
> for something that can run in an xterm while I'm writing email and I
> can see it actually pinging live. The others are snapshots of the
> network, not real time. If it pings every 5 minutes, for 20 seconds,
> and the next 270 seconds the network is down, then comes back, the
> utilities are going to show 100% up. Unless the intermittent
> issue happens during the polling cycle, we won't see it. I'm running
> into the problem of the WISP seeing the graphing show 100% up, while
> I'm sitting here with 64 second chunks of network disappearing. I
> also can't prove its ONLY my site happening, since I am only at this
> one site. I wanted something that I could leave running in a "screen"
> on one of his DD-WRT installed Buffalo (SIGH... That lawsuit sucks)
> routers.
> 
> 		Thanks, Tuc
> _______________________________________________
> cisco-nsp mailing list  cisco-nsp at puck.nether.net
> https://puck.nether.net/mailman/listinfo/cisco-nsp
> archive at http://puck.nether.net/pipermail/cisco-nsp/
> 



More information about the cisco-nsp mailing list