Uploaded image for project: 'Observium'
  1. Observium
  2. OBS-4763

Number of connections for pool members in graph and HTML don't agree

Details

    • Bug
    • Resolution: Not A Bug
    • Minor
    • None
    • None
    • Default
    • None

    Description

      The number of connections on 1 pool member shown in the HTML table for pool members is a much bigger number than the number of current connections shown in the graph for the same table row.

      Attachments

        Activity

          [OBS-4763] Number of connections for pool members in graph and HTML don't agree

          Added "/sec" to graph legend

          adama Adam Armstrong added a comment - Added "/sec" to graph legend

          The two OIDs sound the same, but are conveying different things, the total oid allows us to generate a "connections/sec" number, whereas the "current" OID allows us to display a "current connections" number.

          I assumed this was what was happening as I vaguely remember these devices having these two OIDs for a lot of metrics.

          If you look at the source command for the graph, it's using the "TotConns" DS, so it's graphing connections/sec

          We don't seem to put the instantaneous metric into RRD, I'm not sure why!

          adama Adam Armstrong added a comment - The two OIDs sound the same, but are conveying different things, the total oid allows us to generate a "connections/sec" number, whereas the "current" OID allows us to display a "current connections" number. I assumed this was what was happening as I vaguely remember these devices having these two OIDs for a lot of metrics. If you look at the source command for the graph, it's using the "TotConns" DS, so it's graphing connections/sec We don't seem to put the instantaneous metric into RRD, I'm not sure why!
          nschmalenberger Nick Schmalenberger added a comment - - edited

          I think the HTML table uses ltmPoolMemberStatServerCurConns from mysql, but ltmPoolMemberStatServerTotConns is also polled, is ltmPoolMemberStatServerTotConns used for the RRD input? Maybe this is where the disagreement starts, in the file /opt/observium/includes/polling/loadbalancer/f5-bigip-local-mib.inc.php 

          It makes sense these wouldn't agree, if the connections mostly last longer than a polling period. The delta of ltmPoolMemberStatServerTotConns values would be the number of NEW connections in a polling period, and ltmPoolMemberStatServerCurConns reflects the history of several polling periods for outstanding connections. So connections mostly lasting for 10 polling periods for example would make a 10x difference in these values.

          Both metrics are useful to see, so maybe can this be explained more clearly in the UI, or maybe graph both values?

          nschmalenberger Nick Schmalenberger added a comment - - edited I think the HTML table uses ltmPoolMemberStatServerCurConns from mysql, but ltmPoolMemberStatServerTotConns is also polled, is ltmPoolMemberStatServerTotConns used for the RRD input? Maybe this is where the disagreement starts, in the file /opt/observium/includes/polling/loadbalancer/f5-bigip-local-mib.inc.php  It makes sense these wouldn't agree, if the connections mostly last longer than a polling period. The delta of ltmPoolMemberStatServerTotConns values would be the number of NEW connections in a polling period, and ltmPoolMemberStatServerCurConns reflects the history of several polling periods for outstanding connections. So connections mostly lasting for 10 polling periods for example would make a 10x difference in these values. Both metrics are useful to see, so maybe can this be explained more clearly in the UI, or maybe graph both values?

          F5-BIGIP-LOCAL-MIB::ltmPoolMemberStatServerCurConns."/Common/app-insights-nginx"."/Common/rccp401-22a.iad6.prod.example.com".8080 = Gauge32: 582
          F5-BIGIP-LOCAL-MIB::ltmPoolMemberStatServerCurConns."/Common/app-insights-nginx"."/Common/rccp402-22a.iad6.prod.example.com".8080 = Gauge32: 606
          F5-BIGIP-LOCAL-MIB::ltmPoolMemberStatServerCurConns."/Common/app-insights-nginx"."/Common/rccp403-22a.iad6.prod.example.com".8080 = Gauge32: 576
          F5-BIGIP-LOCAL-MIB::ltmPoolMemberStatServerCurConns."/Common/app-insights-nginx"."/Common/rccp404-22a.iad6.prod.example.com".8080 = Gauge32: 548
          F5-BIGIP-LOCAL-MIB::ltmPoolMemberStatServerCurConns."/Common/app-insights-nginx"."/Common/rccp405-22a.iad6.prod.example.com".8080 = Gauge32: 632
          F5-BIGIP-LOCAL-MIB::ltmPoolMemberStatServerCurConns."/Common/app-insights-nginx"."/Common/rccp406-22a.iad6.prod.example.com".8080 = Gauge32: 592
          F5-BIGIP-LOCAL-MIB::ltmPoolMemberStatServerCurConns."/Common/app-insights-nginx"."/Common/rccp407-22a.iad6.prod.example.com".8080 = Gauge32: 602
          F5-BIGIP-LOCAL-MIB::ltmPoolMemberStatServerCurConns."/Common/app-insights-nginx"."/Common/rccp408-22a.iad6.prod.example.com".8080 = Gauge32: 568

          nschmalenberger Nick Schmalenberger added a comment - F5-BIGIP-LOCAL-MIB::ltmPoolMemberStatServerCurConns."/Common/app-insights-nginx"."/Common/rccp401-22a.iad6.prod.example.com".8080 = Gauge32: 582 F5-BIGIP-LOCAL-MIB::ltmPoolMemberStatServerCurConns."/Common/app-insights-nginx"."/Common/rccp402-22a.iad6.prod.example.com".8080 = Gauge32: 606 F5-BIGIP-LOCAL-MIB::ltmPoolMemberStatServerCurConns."/Common/app-insights-nginx"."/Common/rccp403-22a.iad6.prod.example.com".8080 = Gauge32: 576 F5-BIGIP-LOCAL-MIB::ltmPoolMemberStatServerCurConns."/Common/app-insights-nginx"."/Common/rccp404-22a.iad6.prod.example.com".8080 = Gauge32: 548 F5-BIGIP-LOCAL-MIB::ltmPoolMemberStatServerCurConns."/Common/app-insights-nginx"."/Common/rccp405-22a.iad6.prod.example.com".8080 = Gauge32: 632 F5-BIGIP-LOCAL-MIB::ltmPoolMemberStatServerCurConns."/Common/app-insights-nginx"."/Common/rccp406-22a.iad6.prod.example.com".8080 = Gauge32: 592 F5-BIGIP-LOCAL-MIB::ltmPoolMemberStatServerCurConns."/Common/app-insights-nginx"."/Common/rccp407-22a.iad6.prod.example.com".8080 = Gauge32: 602 F5-BIGIP-LOCAL-MIB::ltmPoolMemberStatServerCurConns."/Common/app-insights-nginx"."/Common/rccp408-22a.iad6.prod.example.com".8080 = Gauge32: 568

          People

            adama Adam Armstrong
            nschmalenberger Nick Schmalenberger
            Votes:
            0 Vote for this issue
            Watchers:
            3 Start watching this issue

            Dates

              Created:
              Updated:
              Resolved: