<html>
<head>
<style><!--
.hmmessage P
{
margin:0px;
padding:0px
}
body.hmmessage
{
font-size: 10pt;
font-family:Tahoma
}
--></style>
</head>
<body class='hmmessage'>You may want to use the bps MIB's instead, such as IfStatsInBitsPerSec and IfStatsOutBitsPerSec. They would be under your vendor's MIB's. They work a lot better and don't wrap. I'm graphing multi-gigabit speeds no problem with them.<br><br>From: rick.jones2@hp.com<br>To: rrd-users@lists.oetiker.ch<br>Date: Tue, 14 Jun 2011 16:52:35 -0700<br>Subject: [rrd-users] bug in dealing with counter wrap in 1.4.3?<br><br><pre>At the moment, I am working with a device that should be using 64-bit<br>wide counters in its report of octets via sFlow, but is only using 32<br>bits. Through this device I am sending a constant stream of netperf<br>UDP_STREAM reporting a very steady 961 Mbit/s (and change, occasionally<br>it will report 960.97 Mbit/s) so I suspect the device is being driven at<br>link-rate. It is reporting these counters every 30 seconds, link-rate 1<br>GbE networking should wrap a 32-bit octet counter in ~40 seconds, so I<br>believe I should never have more than one counter wrap per sampling<br>interval. I am using a step size of one second.<br> <br>I was seeing large gaps in my graphs so I isolated the one value I was<br>looking at from the rest, and captured some data from the device and fed<br>it to an rrd created with:<br> <br>rrdtool create test.rrd --start 1308092201 --step 1 DS:foo:DERIVE:60:0:U<br>RRA:AVERAGE:0.5:1:86400<br> <br>fed it with the data I extracted from a tcpdump trace of the sFlow PDUs:<br> <br>raj@tardy:~/rrdbug$ cat update.sh <br>rrdtool update test.rrd 1308092202.610:441241050<br>rrdtool update test.rrd 1308092232.610:4141751743<br>rrdtool update test.rrd 1308092262.609:3546818048<br>rrdtool update test.rrd 1308092292.609:2952516367<br>rrdtool update test.rrd 1308092322.609:2357287750<br>rrdtool update test.rrd 1308092352.609:1762735341<br>rrdtool update test.rrd 1308092382.609:1168026482<br>rrdtool update test.rrd 1308092412.609:573080395<br>rrdtool update test.rrd 1308092442.609:4273817690<br>rrdtool update test.rrd 1308092472.608:3678028081<br>rrdtool update test.rrd 1308092502.608:3082845004<br>rrdtool update test.rrd 1308092532.608:2488508409<br>rrdtool update test.rrd 1308092562.608:1892937812<br>rrdtool update test.rrd 1308092592.608:1298202985<br>rrdtool update test.rrd 1308092622.608:703006246<br>rrdtool update test.rrd 1308092652.608:107385307<br>rrdtool update test.rrd 1308092682.607:3807847414<br>rrdtool update test.rrd 1308092712.607:3212758463<br>rrdtool update test.rrd 1308092742.607:2617461450<br>rrdtool update test.rrd 1308092772.607:2022995653<br>rrdtool update test.rrd 1308092802.607:1428422068<br>rrdtool update test.rrd 1308092832.607:833109777<br>rrdtool update test.rrd 1308092862.607:239454268<br>rrdtool update test.rrd 1308092892.607:3939282433<br>rrdtool update test.rrd 1308092922.606:3343962564<br>rrdtool update test.rrd 1308092952.606:2749615279<br>rrdtool update test.rrd 1308092982.606:2154394252<br>rrdtool update test.rrd 1308093012.606:1559964909<br>rrdtool update test.rrd 1308093042.606:965307748<br>rrdtool update test.rrd 1308093072.606:370349517<br>rrdtool update test.rrd 1308093102.606:4070642890<br>rrdtool update test.rrd 1308093132.606:3475575191<br>rrdtool update test.rrd 1308093162.605:2881212866<br>rrdtool update test.rrd 1308093192.605:2286174031<br>rrdtool update test.rrd 1308093222.605:1691311682<br>rrdtool update test.rrd 1308093252.605:1097016331<br>rrdtool update test.rrd 1308093282.605:502274162<br>rrdtool update test.rrd 1308093312.605:4201809009<br>rrdtool update test.rrd 1308093342.605:3606613788<br>rrdtool update test.rrd 1308093372.604:3012392733<br>rrdtool update test.rrd 1308093402.604:2417665658<br> <br>and then graphed that:<br> <br>rrdtool graph graph.png --end 1308093373 --start 1308092202 -w 1000 -h<br>400 --step 1 -t "30 Second Counter Samples" -v "Octets/s"<br>--full-size-mode DEF:bar=test.rrd:foo:AVERAGE LINE2:bar#FF0000<br> <br>The resulting graph is attached. Is there perhaps some issue in the<br>wrap code that results in the drop-outs? FWIW, I appear to see the same<br>behaviour if I use COUNTER rather than DERIVE.<br> <br>rick jones<br></pre><br>_______________________________________________
rrd-users mailing list
rrd-users@lists.oetiker.ch
https://lists.oetiker.ch/cgi-bin/listinfo/rrd-users </body>
</html>