Member of The Internet Defense League Últimos cambios
Últimos Cambios
Blog personal: El hilo del laberinto Geocaching

MRTG Rounding or Truncating

Última Actualización: 23 de Noviembre de 1.998 - Lunes 

Resent-Date: Fri, 20 Nov 1998 14:36:49 +0100 (MET)
Old-Return-Path: <jcea@argo.es>
Message-ID: <36557578.FA4953E8@argo.es>
Date: Fri, 20 Nov 1998 12:58:16 -0100
From: Jesús Cea Avión <jcea@argo.es>
Organization: Argo Redes y Servicios Telematicos, S.A.
To: Lista MRTG <mrtg@list.ee.ethz.ch>
Subject: [MRTG] Rounding or truncating?

Since MRTG saves the rate bytes per second in the log file, and uses only integer values, how are decimals managed?. The fractions are rounded or truncated?.

Since I use MRTG for accounting purposes, if it uses truncation, you can have, worst case, the following "deficits":


5 minutes average
30 minutes average
120 minutes average
24 hours average 		Intrinsic
7 bps
40 bps
24 bps
88 bps 		Inherited
-
7 bps
82 bps
352 bps 		Total
7 bps
82 bps
352 bps
4312 bps
(50 hours)
(12.5 days) !!
(50 days) !!
(732 days) !!

I'm not familiar enough with perl to answer this question reading the source code. I'm using MRTG 2.1. The behaviour has changed?. Must I upgrade?. 

-- 
Jesus Cea Avion                         _/_/      _/_/_/        _/_/_/
jcea@argo.es http://www.argo.es/~jcea/ _/_/    _/_/  _/_/    _/_/  _/_/
                                      _/_/    _/_/          _/_/_/_/_/
PGP Key Available at KeyServ   _/_/  _/_/    _/_/          _/_/  _/_/
"Things are not so easy"      _/_/  _/_/    _/_/  _/_/    _/_/  _/_/
"My name is Dump, Core Dump"   _/_/_/        _/_/_/      _/_/  _/_/
"El amor es poner tu felicidad en la felicidad de otro" - Leibniz

--
* To unsubscribe from the mrtg mailing list, send a message with the
  subject: unsubscribe to mrtg-request@list.ee.ethz.ch
* The mailing list archive is at http://www.ee.ethz.ch/~slist/mrtg


Resent-Date: Fri, 20 Nov 1998 21:41:40 +0100 (MET)
Message-ID: <3655DE63.B3B143AA@argo.es>
Date: Fri, 20 Nov 1998 20:25:55 -0100
From: Jesús Cea Avión /lt;jcea@argo.es>
Organization: Argo Redes y Servicios Telematicos, S.A.
To: Lista MRTG 
Subject: Re: [MRTG] Rounding or truncating?
References: <36557578.FA4953E8@argo.es>

> Since MRTG saves the rate bytes per second in the log file, and uses
> only integer values, how are decimals managed?. The fractions are
> rounded or truncated?.

I've studied "rateup.c" from MRTG 2.1 and the answer is: it truncates!!.

Here you have a patch to rounding. It's not checked and must be applied to MRTG 2.1. Not the most current version, you know :-)

The patch is not perfect because MRTG should keep the truncated "last read", not the real "last read". For example:

Time between samples: 300 seconds (5 minutes)

Current
412108
1105330
2005854
2541658
3011190 		Last
0
412108
1105330
2005854
2541658 		Real Bps
1373.6933
2310.74
3001.7466
1786.0133
1565.1066 		Reported
1373
2310
3001
1786
1565 		Reported Bps (1)
1374
2311
3002
1786
1565 		Reported Bps (2)
1373
2311
3002
1786
1565
Bytes Missed 690 -210 90 (3)

  1. This patch (rounding)

  2. A patch with rounding for 30 min, 120 min and 24 hours, like (1), but keeping "multiple of the sample-time" in log, instead of the real in/out counters.

    Real Counter
    450
    890
    1190
    1658     		Value stored
    300
    600
    900
    1500

    The graph for the current MRTG would be: 

        450, 440 (850-450), 300 (1190-890), 468 (1658-1190)
    Dividing by 300 
        1, 1, 1, 1
    Using the "multiple of the sample-time", the graph would be: 
        1, 1, 1, 2
    That is, the bytes missed in each sample are accumulated. So, they are not lost. Only "delayed".

    The problem with such a patch is that "last sample" is stored in the log as a string, not as a number. So, if we want to modify it we must add a routine to do so, like the "diff()" routine.

    We can't either to modify the MRTG perl because it has no time references.

    Any idea?. Are these patches applicable to current MRTG release?.

  3. The bytes missed are bounded to "sample-time". 300 in this example. So, in the long run, the error is ALWAYS bounded to 300.

The following patch is (1), not (2): 


--- rateup.c.old        Fri Nov 20 17:58:39 1998
+++ rateup.c    Fri Nov 20 18:36:04 1998
@@ -135,7 +135,7 @@
        outmax = max(outmax,history[n].outmax);\
        if (now < history[n].time && n < histvalid) n++; \
     } while(now < history[n].time && n < histvalid) ; \
-    inr /= avc; outr /= avc; now += steptime;
+    inr=(inr+avc/2)/avc; outr=(outr+avc/2)/avc; now += steptime;
 
 
 
@@ -660,8 +660,8 @@
            inrate = diff(in,last.in);
            outrate = diff(out,last.out);
        } else {
-           inrate = diff(in,last.in) / period;
-           outrate = diff(out,last.out) / period;
+           inrate = (diff(in,last.in)+period/2) / period;
+           outrate = (diff(out,last.out)+period/2) / period;
        }
     }
     if (inrate < 0 || inrate > abs_max) inrate = history[0].in;



-- 
Jesus Cea Avion                         _/_/      _/_/_/        _/_/_/
jcea@argo.es http://www.argo.es/~jcea/ _/_/    _/_/  _/_/    _/_/  _/_/
                                      _/_/    _/_/          _/_/_/_/_/
PGP Key Available at KeyServ   _/_/  _/_/    _/_/          _/_/  _/_/
"Things are not so easy"      _/_/  _/_/    _/_/  _/_/    _/_/  _/_/
"My name is Dump, Core Dump"   _/_/_/        _/_/_/      _/_/  _/_/
"El amor es poner tu felicidad en la felicidad de otro" - Leibniz

--
* To unsubscribe from the mrtg mailing list, send a message with the
  subject: unsubscribe to mrtg-request@list.ee.ethz.ch
* The mailing list archive is at http://www.ee.ethz.ch/~slist/mrtg

Python Zope ©1998 jcea@jcea.es

Más información sobre los OpenBadges

Donación BitCoin: 19niBN42ac2pqDQFx6GJZxry2JQSFvwAfS