Hi Alexei, I made the changes to all the files and everything works again as before but it is prettier this way. Thank you for your help!
However, my initial issue is still there: on the "Details" tab, when I change from minute to minute using the "browsing" buttons (I don't know how to call them) located right below the graphs, I see it moves backward or forward on the graph by one minute and the graph shows different level of traffic/packets/flows and the minute displayed at tstart and tend change, but the value in the "Statistics" section below the graph does not reflect this and displays the same value for every time slots of 5 minutes, displaying the value of the first minute for all the next four minutes. It has this behavior for Shadow profiles only, with either "1:1 channels from profile live" or "individual channels". Real Profiles are fine. I checked the various files in libexec but since I have no knowledge of Perl, I got lost at some point trying to figure out where the value displayed in the "Statistics" section came from. My understanding was that values for the graph and for the "Statistics" came from the same RRD, so I'm puzzled by the discrepancy. I value your input on this issue as I might have missed something in my NFSen configuration to end up with this behavior. Thank you for your advice. Regards, Yannick -----Original Message----- From: Alexei Pastuchov [mailto:[email protected]] Sent: Thursday, June 09, 2016 4:43 AM To: Yannick D'Amboise; [email protected] Subject: RE: [Nfsen-discuss] Unexact results in Shadow profile Statistics timeslot after changing CYCLETIME value Hi Yannick, 1. RRDs::create --step argument in SetupAlertRRD is required as well. It's added in https://github.com/p-alik/nfsen/commit/781cbf95437f9e2200965c715bee582cecce25b2 2. An alternative to your hard coded solution could be done in this way: https://github.com/p-alik/nfsen/tree/variable-rrd-heartbeat It works but I don't how valuable the change is. Therefor I put it in separate branch. Cheers, Alexei > On 08 June 2016 at 22:02 Yannick D'Amboise <[email protected]> wrote: > > > Hi Alexei, > > Thank you very much for taking some of your time to make those changes > and pointing me to them. > > I will test those changes and will let you know how it goes. > > In the meantime, I have two questions: > > 1. In NfSenRRD.pm, the change made to line 121 (adding "--step", > $NfConf::CYCLETIME) is not repeated on line 174 which has the exact same code: > when I hardcoded the value for step, I did it on both lines and I > wonder if I had had gone too far by doing this. Your input (or anyone > else's) is more than welcome on this. > 2. Also in NfSenRRD.pm on line 99 and 160 (line numbers might vary slightly), > I had hardcoded the heartbeat value in push @DS, > "DS:$ds:ABSOLUTE:600:U:U"; to push @DS, "DS:$ds:ABSOLUTE:120:U:U"; > . It might be useless or undesirable: I value your input and > advice in advance regarding this as well . > > Thank you again for these changes that go far beyond what I had hardcoded. > > Regards, > > Yannick > > > -----Original Message----- > From: Alexei Pastuchov [mailto:[email protected]] > Sent: Wednesday, June 08, 2016 10:26 AM > To: Yannick D'Amboise; [email protected] > Subject: Re: [Nfsen-discuss] Unexact results in Shadow profile > Statistics timeslot after changing CYCLETIME value > > Hi Yannick, > CYCLETIME is an issue in my case too. > So I patched nfsen implementation and actually graphs in unpatched and > patched versions look absolutely equal. > I think this line could contains a solution you are looking for. > https://github.com/p-alik/nfsen/blob/master/libexec/NfSenRRD.pm#L121 > > Cheers, > Alexei > > > On 02 June 2016 at 22:55 Yannick D'Amboise > > <[email protected]> > > wrote: > > > > > > Hi, > > > > As added details, here is my set-up: > > > > - OS: CentOS 7 > > - NFSEN: nfsen-1.3.7 > > - NFDump: nfdump-1.6.13 > > > > I've googled about my issue and found very few things; I've searched > > the mailing list and found only something regarding value that are > > not exact when looking at results in the past but this is expected > > as data is being averaged the further back in time they are. I also > > found a post regarding changing polling interval stating that > > changing CYCLETIME should mostly do it and that NFSen 1.3.7 should > > already be flexible (I'm running 1.3.7). > > > > I've changed all the instance of the hard-coded "300"including > > CYCLETIME value I could find in the in the NFSen directory and > > looked for instances under other possible names without success; if > > anyone can point me to something I missed, I'd be more than happy. > > > > I created a XML file of the content of the RRD using RRDDump to look > > at the data itself and could confirm that there was a variation > > every > > 60 seconds so my guess is that the config changes I made to > > NfSenRRD.pm were correct; there probably is a better way of getting > > data from the database but I'm new to RRD so please forgive my ignorance. > > > > I also had a look at the logs (/var/log/messages) and I see a lot of > > NFSen NFCapd, etc messages, but no errors. > > > > Please note that I have no experience with Perl nor PHP, my > > background in coding helps me a bit when looking at the code but is > > somewhat limited also so I might have missed something obvious. I've > > been through details.php (in the NFSen WWW folder) and could trace a > > few things but my limited knowledge stops me at some points. I get > > to the following line (1203) in the file: > > > > $_val = ScaleBytes($statinfo['traffic' . $_type], $rateval, 1000.0); > > > > But then I get lost when tracking "$rateval". I find this > > > > $rateval = $_SESSION['tright'] - $_SESSION['tleft'] + $CYCLETIME; > > > > At line 1139, but I'm not able to interpret it. Any hint would be welcome. > > > > Obviously, shadow profiles are really useful as they eat less disk > > space and that's why I'm trying to make this work. > > > > English not being my native language, please excuse any ambiguity. > > > > Thank you very much for your help and support. Any and all help is > > deeply appreciated. > > > > Regards, > > > > Yannick > > > > > > -------------------------------------------------------------------- > > -- > > -------- What NetFlow Analyzer can do for you? Monitors network > > bandwidth and traffic patterns at an interface-level. Reveals which > > users, apps, and protocols are consuming the most bandwidth. > > Provides multi-vendor support for NetFlow, J-Flow, sFlow and other > > flows. Make informed decisions using capacity planning reports. > > https://ad.doubleclick.net/ddm/clk/305295220;132659582;e > > _______________________________________________ > > Nfsen-discuss mailing list > > [email protected] > > https://lists.sourceforge.net/lists/listinfo/nfsen-discuss ------------------------------------------------------------------------------ What NetFlow Analyzer can do for you? Monitors network bandwidth and traffic patterns at an interface-level. Reveals which users, apps, and protocols are consuming the most bandwidth. Provides multi-vendor support for NetFlow, J-Flow, sFlow and other flows. Make informed decisions using capacity planning reports. https://ad.doubleclick.net/ddm/clk/305295220;132659582;e _______________________________________________ Nfsen-discuss mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/nfsen-discuss
