-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Hi Adrian,

- -------- Original Message  --------
From: Adrian Popa <[EMAIL PROTECTED]>
To: [email protected]
Subject: [Nfsen-discuss] Give a man enough rope...
Date: Wed Oct 04 2006 10:37:17 GMT+0200 (CEST)

> Hello,
> You probably know the saying: Give a man enough rope and he will hang
> himself. Well, open-source software is sometimes known to 'give enough
> rope'. I'm sorry to say that the same thing can't be said about nfdump. 
> :-)
> 
> I tried to do a top 10000 (yes, it's big, but that's my problem,
> right?), and nfdump wouldn't let me because a valid topN is between 0
> and 1000. So, I set out to remove this barrier, and I managed to do it.
> 
> in nfdump.c (version 1.5.2), at line 1144 I modfied from:
> if ((aggregate || flow_stat)  && ( topN > 1000 || topN == 0) ) {
> to
> if ((aggregate || flow_stat)  && ( topN > 10000 || topN == 0) ) {
> 
> I recompiled everything, and as magic, it works! (I admit, I was very
> surprised that it was so easy to do...)  :-) So, you can say that I
> manufactured my extra rope... :)
> 
> My question would be: why is this limitation imposed? Because everything
> seems to work fine with a greater top...

You're right, that the only limit is that line. You may do top 
'what_ever_you_want' from the code perspective, as the code does everything 
dynamically. So way this limitation? This magic boundary was
introduced because of speed reason. nfdump uses 2 different sort algorithms for 
record statistics and record element statistics. You may have noticed, that you 
can to -n 0 for -s ip, which results in
a fully sorted list of IP addresses from your flows. Sorting records uses 
another algorithm, which is faster for average top N usage ( typically 10 - 100 
), but slower > 1000 and can become very slow
for a full sort ( -n 0 ). That's why the limit was set to 1000. So the rope was 
limited - typically long enough, but too short to hang yourself :) So you see 
I'm afraid of you health.
You may remove any restriction - but don't complain, if the rope gets long 
enough.

Maybe one day I will change that to dynamically selecting the proper faster 
algorithm, according to -n .

Hope this helps.

Take care.

        - Peter


> 
> Sorry for my ranting,
> Adrian Popa
> 
> 
> ------------------------------------------------------------------------
> 
> -------------------------------------------------------------------------
> Take Surveys. Earn Cash. Influence the Future of IT
> Join SourceForge.net's Techsay panel and you'll get the chance to share your
> opinions on IT & business topics through brief surveys -- and earn cash
> http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV
> 
> 
> ------------------------------------------------------------------------
> 
> _______________________________________________
> Nfsen-discuss mailing list
> [email protected]
> https://lists.sourceforge.net/lists/listinfo/nfsen-discuss

- --
_______ SWITCH - The Swiss Education and Research Network ______
Peter Haag,  Security Engineer,  Member of SWITCH CERT
PGP fingerprint: D9 31 D5 83 03 95 68 BA  FB 84 CA 94 AB FC 5D D7
SWITCH,  Limmatquai 138,  CH-8001 Zurich,  Switzerland
E-mail: [EMAIL PROTECTED] Web: http://www.switch.ch/security
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.3 (Darwin)

iQCVAwUBRSN6g/5AbZRALNr/AQIBbwP/ZLNrTcRRntruDlvOzkBR3wD8U0E4/4P0
ZrSKpG8oCMcQBlldi6TJnjReOCF+tzA0U5MZQYjejFaq0RmY3BNUOwxeDAAysplp
cAvjN/Rxprh0kTwPgAhMIzIVgqS19B46IIG0j/y0m3y2DZCLOmrMBfP3lU4ZK1Vp
47u+XEGMDvo=
=ilHy
-----END PGP SIGNATURE-----

-------------------------------------------------------------------------
Take Surveys. Earn Cash. Influence the Future of IT
Join SourceForge.net's Techsay panel and you'll get the chance to share your
opinions on IT & business topics through brief surveys -- and earn cash
http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV
_______________________________________________
Nfsen-discuss mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/nfsen-discuss

Reply via email to