RRD incorrect traffic totals [resolved]



  • Looking at the pictures below - something is obviously not right.  Downloading at 15 Mb/sec for over 2 hours, I'd downloaded over 13 GB of data (verified by checking how much I actually had downloading, and via calculator), yet the graph summary shows only around 500 MB of data received.  This incorrectness exists in all the period charts.  Also, even though the graphs only have data going back about 2 weeks, the 1 month, 6 month, and 1 year graph all show more data downloaded than the previous (in age) graph.  Is anyone else seeing this anomaly?  If there is any further data I can provide, please let me know.

    Currently running the March 3 afternoon build, but have seen this error for at least 1 week.









  • reset your RRD graphs

    http://forum.pfsense.org/index.php/topic,1748.msg10114.html#msg10114

    Download something good in size, like vmware player (they have fast downloads)

    Keep in mind that the initial graph that you attached only covers a 4 hour period.

    Reply if they are off again.



  • That link while good in describing how to reset the RRD graphs, does not solve my problem.  Disabling RRD, deleting its logs, enabling it again, and downloading a known good 2 GB file still yields incorrect totals.  Each graph shows a 15 Mb/sec download, but with each chart showing a totally wrong numbers (4 hours = 8 MB, 16 hours = 10 MB, and the rest showing 0).

    Further, if you look at the average numbers for in and out vs the total vs in and out, they just don't make sense…the same is true even after deleting all RRD data.  I mean, the average on the outbound is 54 times lower, yet the the total data is only 1.5 times lower (on the below 4 hour graph...and similarly on the graph generated by my fresh RRD data).  Something somewhere is clearly not right.

    Are the numbers and graphs on your RRD traffic tab all sane?



  • Yes. I haven't been able to replicate this issue on my test install.

    Are you using the latest beta?



  • Just tried with the latest snapshot, Friday Mar 5 06:17:42 EST 2010 and still have the same issue.  Updated to snapshot, cleared RRD graph data, downloaded 1.2 GB.  The graph totals show, 67 MB, 64 MB, 31 MB, and the last 4 all show 0 bytes (their period has not yet elapsed, so likely normal).

    Certainly an odd issue and one which I never had in 1.2.3. One thing of note, is I am running this on vmware server, but I was running 1.2.3 the same way without any difficulty.



  • Found the problem.  I noticed in my logs, I had this error:

    php: : The command '/usr/bin/nice -n20 /usr/local/bin/rrdtool update /var/db/rrd/wan-queues.rrd -t :wan:qACK:qDefault:qP2P:qVoIP:qGames:qOthersHigh:qOthersLow N:U:U:U:U:U:U:U:U' returned exit code '1', the output was 'ERROR: unknown DS name '''
    

    Removed the traffic shaper, ran rm /var/db/rrd/queue and re-enabled the traffic shaper and now everything is grand.  As expected, I've lost all queue data, but the totals are there and now make sense.  As a good side effect, my status -> queues also now make sense (before they did not show data that made sense).

    The problem does come back quickly though.


Log in to reply