The value from the stats is clearly worth it and I can afford the bandwidth just fine (now that ntopSupport.com is moved over, I have 5 GB/month with 1and1). Really, it's a teeny tiny load. With the 3.0 release we moved the primary to Luca's server, with mine as the backup - so all I'm seeing are the people still running old versions (2.2.98 -> rc1) and an occasional fallover.
It's just that the one user was creating so many records (6k of the 15k records in my logs, 14% of the total) that it was distorting the statistics. Although whomever it is is still anonymous, I sent the msg at 11:49 AM (US CST), and it stopped 41 minutes later... :-) I added the ability to block a specific IP for a given time range and the reports are now clean. But seriously, look at the stats http://www.ntopsupport.com/alr.html. What's it worth as a developer to know * That usage of Debian 3.0 and Red Hat 9 are running neck and neck, that Fedora Core is 1/2 of those counts and nothing else is close? * That usage of Linux is running almost 10:1 vs. FreeBSD (evenly split 4 vs 5) and 13:1 vs Windows? 10822 Linux 82.6% 1065 FreeBSD4/5 8.1% 843 Windows 6.4% 549 FreeBSD4 4.2% 516 FreeBSD5 3.9% 255 Solaris 1.9% 78 Darwin 0.6% 18 Unknown 0.1% 14 OpenBSD 0.1% * That there are still 80+ people running GCC 2.91? (Get with the 00s folks) * That most people are still using zlib 1.1.4, but almost 32% are using 1.2.x?? etc. Are these Scientific samples? No. * The person who has bravely tried every version from 2.2.98, 2.2.99, 3.0pre1, 3.0pre2, 3.0rc1 and now 3.0 counts 6x vs the new 3.0 user. (Over time this will correct itself, or I'll change the reporting routines to drop versions < 3.0) * There are undercounts due to NAT. But not as many as you thing - unless the machines are EXACTLY the same, I see them as different. * There are undercounts due to people opting out of data collection. * Maybe the undercounts aren't evenly distributed - But if even 10x as many OpenBSD users are paranoid and opt-out, it's still a small population. I can live with these limits. It's still data with a lot of value. -----Burton > -----Original Message----- > From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] Behalf Of > Horta, Benny > Sent: Tuesday, March 30, 2004 12:28 PM > To: '[EMAIL PROTECTED]'; Ntop-Dev > Subject: RE: [Ntop] Does anybody recognize this ip: 64.202.160.1xx? > > > Looks like its time to buy a bigger pipe or disable by default autoversion > checking. I think its probably safer to disable automatic version checking > for the end user. > > Like the old saying give them an inch and they take the whole yard. Or it > takes one bad apple to spoil the bunch. > > > -----Original Message----- > From: Burton M. Strauss III [mailto:[EMAIL PROTECTED] > Sent: Tuesday, March 30, 2004 12:49 PM > To: Ntop; Ntop-Dev > Subject: [Ntop] Does anybody recognize this ip: 64.202.160.1xx? > > You are running an ntop instance (3.0pre1) that is restarting > every minute. > > Please fix it! > > -----Burton > > > > The name resolves back to a useless address: > ip-64-202-160-1xx.secureserver.net adn the registrant is GoDaddy, > so that's > just one of the big hosting/registry cos. > > Here's ONE of the 800 log records from today: > > 64.202.160.1xx - - [30/Mar/2004:02:21:13 -0500] "GET /version.xml > HTTP/1.0" > 200 1574 www.burtonstrauss.com "-" "ntop/3.0pre1+SourceForge+RPM > host/i686-pc-linux-gnu distro/redhat release/9 kernrlse/2.4.20-6smp > GCC/3.2.2 config(...) gdbm/1.8.0 gd/1.8.x openssl/0.9.7a zlib/1.1.4 > access/both interfaces(eth0)" "-" > > > > > > _______________________________________________ > Ntop mailing list > [EMAIL PROTECTED] > http://listgateway.unipi.it/mailman/listinfo/ntop > _______________________________________________ Ntop mailing list [EMAIL PROTECTED] http://listgateway.unipi.it/mailman/listinfo/ntop
