Uploaded image for project: 'Observium'
  1. Observium
  2. OBS-2901

graph.php showing zero agregate bps value for now in multi-port_bits* graphs

Details

    • Bug
    • Resolution: Fixed
    • Major
    • None
    • Professional Edition
    • API
    • None

    Description

      eg. /graph.php?type=multi-port_bits_duo_separate&id=26475,26299 shows zero agregate bps for "now"

      Output attached

      Attachments

        Activity

          [OBS-2901] graph.php showing zero agregate bps value for now in multi-port_bits* graphs

          It's working as it should. Thank you again.

          zidek@master.cz Martin Zídek added a comment - It's working as it should. Thank you again.

          Sorry , I will check it. Thank you.

          zidek@master.cz Martin Zídek added a comment - Sorry , I will check it. Thank you.
          adama Adam Armstrong added a comment - This is definitely fixed. http://demo.observium.org/graph.php?type=multi-port_bits_duo_separate&id=1072&idb=1173

          I performed update  to 9701 and I see your fix with ADDNAN, but aggregate graphs still have zero value for Now. 

          zidek@master.cz Martin Zídek added a comment - I performed update  to 9701 and I see your fix with ADDNAN, but aggregate graphs still have zero value for Now. 

          This was apparently added in 2008, 2 years after i wrote the first graphs!

          adama Adam Armstrong added a comment - This was apparently added in 2008, 2 years after i wrote the first graphs!
          adama Adam Armstrong added a comment - - edited

          Aha. Apparently there is a special operator for this, because 1 + UNKNOWN = UNKNOWN, so they have a special operator where 1 + UNKNOWN = 1.

          It should behave better fin 9701.

          (this is literally a 12 year mystery solved!)

          adama Adam Armstrong added a comment - - edited Aha. Apparently there is a special operator for this, because 1 + UNKNOWN = UNKNOWN, so they have a special operator where 1 + UNKNOWN = 1. It should behave better fin 9701. (this is literally a 12 year mystery solved!)

          I don't think it's even a bug, I think it's a limitation in how rrd stores data. We're showing the "LAST" value of the aggregate of IN+OUT, and it seems that for half of the time period, this isn't populated and returns zero (because the data hasn't been polled yet).

          I'm not sure if there's some strategy to counter this. I recall fiddling with it a few times.

          Looking at it further, it's because we set "unknown" values in the rrd data to zero when graphing, because otherwise any unknown value in the data completely breaks the aggregate (a single down port in a graph of 100 will set all values to unknown).

           

          adama Adam Armstrong added a comment - I don't think it's even a bug, I think it's a limitation in how rrd stores data. We're showing the "LAST" value of the aggregate of IN+OUT, and it seems that for half of the time period, this isn't populated and returns zero (because the data hasn't been polled yet). I'm not sure if there's some strategy to counter this. I recall fiddling with it a few times. Looking at it further, it's because we set "unknown" values in the rrd data to zero when graphing, because otherwise any unknown value in the data completely breaks the aggregate (a single down port in a graph of 100 will set all values to unknown).  

          I just tested it with lastest RRDtool 1.7.1 / Compiled Feb 11 2019 19:34:18 with same result, 0 on aggregate.

          zidek@master.cz Martin Zídek added a comment - I just tested it with lastest RRDtool 1.7.1 / Compiled Feb 11 2019 19:34:18 with same result, 0 on aggregate.

          So this is only problem of aggregate graphs and it's caused by some bug in rrdtool. Have you tried to report this upstream ?

          zidek@master.cz Martin Zídek added a comment - So this is only problem of aggregate graphs and it's caused by some bug in rrdtool. Have you tried to report this upstream ?
          adama Adam Armstrong added a comment - - edited

          This is caused by rrdtool being useless. I'm not sure of the exact reason, but I think it's because it is wanting to aggregate the NEXT period instead of the current one.

          It's been annoying for many years, and we don't find a solution for it.

          adama Adam Armstrong added a comment - - edited This is caused by rrdtool being useless. I'm not sure of the exact reason, but I think it's because it is wanting to aggregate the NEXT period instead of the current one. It's been annoying for many years, and we don't find a solution for it.

          People

            adama Adam Armstrong
            zidek@master.cz Martin Zídek
            Votes:
            0 Vote for this issue
            Watchers:
            2 Start watching this issue

            Dates

              Created:
              Updated:
              Resolved: