Uploaded image for project: 'Network Observability'
  1. Network Observability
  2. NETOBSERV-1955

Wrong metrics when conversation tracking is used

    • Icon: Bug Bug
    • Resolution: Done
    • Icon: Normal Normal
    • netobserv-1.9
    • netobserv-1.7
    • None
    • Quality / Stability / Reliability
    • False
    • Hide

      None

      Show
      None
    • None
    • None
    • None
    • None
    • NetObserv - Sprint 264, NetObserv - Sprint 265, NetObserv - Sprint 266, NetObserv - Sprint 267, NetObserv - Sprint 268
    • None
    • None
    • Hide
      Previously, when using conversation tracking (`logTypes: Conversations` or `logTypes: All` in FlowCollector), the Traffic rates metrics visible in the dashboards were flawed, wrongly showing an seemingly out-of-control increase in traffic. The metrics are now fixed to show more accurate traffic rates. However, note that in `Conversations` and `EndedConversations` modes, these metrics are still not 100% accurate as they don't include long-standing connections. This information has been added to the documentation. The default mode `logTypes: Flows`, is recommended to avoid this kind of inaccuracy.
      Show
      Previously, when using conversation tracking (`logTypes: Conversations` or `logTypes: All` in FlowCollector), the Traffic rates metrics visible in the dashboards were flawed, wrongly showing an seemingly out-of-control increase in traffic. The metrics are now fixed to show more accurate traffic rates. However, note that in `Conversations` and `EndedConversations` modes, these metrics are still not 100% accurate as they don't include long-standing connections. This information has been added to the documentation. The default mode `logTypes: Flows`, is recommended to avoid this kind of inaccuracy.

      When Conversation Tracking is enabled with logTypes=All or logTypes=Conversations, metrics generated (such as bytes and packet rates) show much bigger numbers as expected. This doesn't happen with logTypes=EndedConversations.

      cf https://issues.redhat.com/browse/NETOBSERV-1922?focusedId=25741876&page=com.atlassian.jira.plugin.system.issuetabpanels%3Acomment-tabpanel#comment-25741876

      > "Moreover there seems to be a bug in metrics generation, which probably counts all flows regardless of their log types, ending up in duplicated counts."

       

              jtakvori Joel Takvorian
              jtakvori Joel Takvorian
              None
              None
              None
              Amogh Rameshappa Devapura Amogh Rameshappa Devapura
              None
              Votes:
              0 Vote for this issue
              Watchers:
              2 Start watching this issue

                Created:
                Updated:
                Resolved: