[hobbit] Hobbit CONN questions

Mon Nov 26 00:06:31 CET 2007

Why don't you try a tcpdump or snoop against a specific host and see what is going on?  I think you are being misled by these measurements.

If I remember the code, the Hobbit pinger tries once every minute.  It also will retry a few times if there is a failure.  If it decides to put "conn" into a red state, it then starts polling every minute (instead of five minutes) for 30 minutes.  I think this was done to shorten the down times for ping tests.  After 30 minutes, the pinger reverts to its normal cycle.

GLH

________________________________

	From: Josh Luthman [mailto:josh at imaginenetworksllc.com] 
	Sent: Thursday, November 22, 2007 6:18 PM
	To: hobbit at hswn.dk
	Subject: Re: RE: Re: [hobbit] Hobbit CONN questions

	Two hints:

	1) When a host goes red, it always goes green after 28-32 seconds or 43-46 seconds.
	2) When a host is red divide the number of seconds it has been unchanged by the amount of polls.  You'll never get a a remainder greater then a minute. 

	On 11/22/07, Hubbard, Greg L <greg.hubbard at eds.com> wrote: 

		"Hobbit polls every 30 - 45 seconds"  -- really?  How did you measure this?

________________________________

			From: Josh Luthman [mailto:josh at imaginenetworksllc.com] 
			Sent: Thursday, November 22, 2007 6:03 PM 

			To: hobbit at hswn.dk 
			Subject: Re: RE: Re: [hobbit] Hobbit CONN questions

			ARP can't be the problem - ARP will cache the result for 2 minutes and clear it if unused.  If the result is used and is valid again within the 2 minutes, it is cached for 10 minutes.  Hobbit polls every 30-45 seconds so ARP is not a problem. 

			Having switch from using hobbitping to fping I think the graphs are going to look far superior now.  Things on the same switch are no longer 30ms but 1ms =)

			I now understand why there are warnings not to use hobbitping, but I don't understand how the command line is getting such normal results while the graphs are getting such high results.  Does anyone know why this is? 

			Josh

			On 11/22/07, Hubbard, Greg L <greg.hubbard at eds.com> wrote: 

				Don't forget the ARP protocol -- if a destination is not in the local system's ARP cache, the lower layer code in the IP stack discards the first packet and generates an ARP broadcast.  The assumption (I guess) is that the discarded packet will be retransmitted.  You might look for the "knobs" on fping to control timeouts and retry counts.

				GLH

________________________________

					From: Josh Luthman [mailto:josh at imaginenetworksllc.com] 
					Sent: Wednesday, November 21, 2007 1:13 AM
					To: hobbit at hswn.dk
					Subject: Re: RE: Re: [hobbit] Hobbit CONN questions

					Concerning fping options, I have some RRD graphs that show enormous pings to some of our customer's (upwards of 800ms) because as the radio doesn't pass any traffic it sits there and when someone first looks for it, it acts kind of dumb. 

					What delegates how often the FPING command is issued?  I'm not looking for the arguments for fping itself, but rather how hobbitd kicks fping into gear.

					I would like to increase the rate of the pings and see if that helps.  I'm running a continuous ping over night to see if that changes the graph or if it really does spike to that outrageous height. 

					Josh

					On 11/9/07, johan.boye at latecoere.fr < johan.boye at latecoere.fr > wrote: 

					> -----Message d'origine-----
					> De : Henrik Stoerner [mailto: henrik at hswn.dk]
					> Envoyé : vendredi 9 novembre 2007 08:23
					> À : hobbit at hswn.dk
					> Objet : Re: RE: Re: [hobbit] Hobbit CONN questions 
					>
					> On Fri, Nov 09, 2007 at 08:14:32AM +0100,
					> johan.boye at latecoere.fr wrote:
					> > I'm just wondering what triggers in the Hobbit code is used
					> to make a RED ping status :
					> >   - after just one no echo-reply each 5min ?
					> >   - after a couple of no echo-reply ? How much ?
					> >   - What is the timeout used to declare the host unreachable ? 
					>
					> It goes red when the ping utility that is used (fping or hobbitping)
					> says that the host did not respond. How many pings are sent
					> and what the
					> timeouts are is decided by the ping tool; see their man-pages for 
					> details. You can tune this through commandline options for the ping
					> tool, these go into the FPINGCMD setting in hobbitserver.cfg
					>
					>
					> Henrik

					Got it, options must be put in the hobbitserver.cfg . Thanks for the info!

					"Les informations contenues dans ce message électronique peuvent être de nature confidentielles et soumises à une obligation de secret. Elles sont destinées à l'usage exclusif du réel destinataire. Si vous n'êtes pas le réel destinataire, ou si vous recevez ce message par erreur, merci de le détruire immédiatement et de le notifier à son émetteur." 

					"The information contained in this e-mail may be privileged and confidential. It is intended for the exclusive use of the designated recipients named above. If you are not the intended recipient or if you receive this e-mail in error, please delete it and immediately notify the sender." 

					To unsubscribe from the hobbit list, send an e-mail to
					hobbit-unsubscribe at hswn.dk

					-- 
					Josh Luthman
					Office: 937-552-2340
					Direct: 937-552-2343
					1100 Wayne St
					Suite 1337
					Troy, OH 45373

					Those who don't understand UNIX are condemned to reinvent it, poorly.
					--- Henry Spencer 

			-- 
			Josh Luthman
			Office: 937-552-2340
			Direct: 937-552-2343
			1100 Wayne St
			Suite 1337
			Troy, OH 45373

			Those who don't understand UNIX are condemned to reinvent it, poorly. 
			--- Henry Spencer 

	-- 
	Josh Luthman
	Office: 937-552-2340
	Direct: 937-552-2343
	1100 Wayne St
	Suite 1337
	Troy, OH 45373

	Those who don't understand UNIX are condemned to reinvent it, poorly. 
	--- Henry Spencer 

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.xymon.com/pipermail/xymon/attachments/20071125/89062f9c/attachment.html>