[smokeping-users] Re: Trouble recording huge pings

Vinny Abello vinny at tellurian.com
Sun Jul 3 20:57:44 MEST 2005

At 02:52 PM 7/3/2005, Aaron Wolfe wrote:
>I am trying to convince my ISP that a router in their network is the
>source of trouble we are having by using smokeping from each side to
>show where the latency happens.
>The problem we are trying to fix is strange, it is very intermittant
>and when the latency occurs, it will last for only 5 to 15 seconds.
>Ping times during these periods go from the normal 40-50ms to as high
>as 15,000ms  (yes, 15,000).   While a command line ping running will
>show these insane values, fping times out at 500ms.  This seems to
>leave holes in the graph with kind of show the problem on very short
>time period graphs but the holes are lost in the longer term graphs
>and so we have a flat line where we should really have huge spikes.

Be aware that these spikes in latency pinging the router itself may 
have nothing to do with traffic flowing through the router. 
Responding to pings and routing pings are two completely different 
things to a router. That's why you might see latency on traceroutes 
which is artificial. Often times when a router CPU is busy, the ping 
times directed at the router will go up because responding to an ICMP 
request is typically the lowest possible priority for it. I observe 
this frequently on my Cisco routers when the BGP scanner process 
kicks in for several seconds. I've never seen the response go 
anywhere near as high as you are saying, but I wanted to throw this 
out there anyway.

Are you seeing the same latency to something beyond the router at the 
exact same time? That's the key to determining if it's really a delay 
at the router or not.

>I have tried pinging 1 packet every second and also 3 packets every 3
>seconds (i believe fping waits 1 second between pings to the same host
>anyway).  I have edited FPing.pm to pass the argument -t15000 to
>fping, hopefully causing fping to wait long enough for the returned
>packets.  My graphs now show some returns at over 1,000 but they more
>often just leave periods of time blank.
>Is there a way to make these times when all packets are timed out look
>really obvious on a long term graph, or a way to get smokeping
>actually recording the HUGE return times I'm seeing?
>Thanks for any help.  These guys are being real PITA about fixing
>their network, insisting the problem is on our end (I have done
>everything under the sun to make sure that is not the case).   Running
>a ping at the commandline shows that it's obviously a particular set
>of routers in their network that the problem occurs, but I need
>something prettier to convince the non technical folks to take action
>Unsubscribe mailto:smokeping-users-request at list.ee.ethz.ch?subject=unsubscribe
>Help        mailto:smokeping-users-request at list.ee.ethz.ch?subject=help
>Archive     http://lists.ee.ethz.ch/smokeping-users
>WebAdmin    http://lists.ee.ethz.ch/lsg2.cgi

Vinny Abello
Network Engineer
Server Management
vinny at tellurian.com
(973)300-9211 x 125
(973)940-6125 (Direct)
PGP Key Fingerprint: 3BC5 9A48 FC78 03D3 82E0  E935 5325 FBCB 0100 977A

Tellurian Networks - The Ultimate Internet Connection
http://www.tellurian.com (888)TELLURIAN

"Courage is resistance to fear, mastery of fear - not absence of 
fear" -- Mark Twain

Unsubscribe mailto:smokeping-users-request at list.ee.ethz.ch?subject=unsubscribe
Help        mailto:smokeping-users-request at list.ee.ethz.ch?subject=help
Archive     http://lists.ee.ethz.ch/smokeping-users
WebAdmin    http://lists.ee.ethz.ch/lsg2.cgi

More information about the smokeping-users mailing list