[jira] [Updated] (NIFI-3411) Latest "NiFi" Status History data point should only be reported when complete

["Joseph Gresock (JIRA)" <ji...@apache.org>]

     [ https://issues.apache.org/jira/browse/NIFI-3411?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

Joseph Gresock updated NIFI-3411:
---------------------------------
    Attachment: Status History 2.png
                Status History 1.png

Status History 1 shows the dip in the graph, and Status History 2 shows the dip resolved after I refreshed the view several times.  Also note the difference in average cluster metrics.

> Latest "NiFi" Status History data point should only be reported when complete
> -----------------------------------------------------------------------------
>
>                 Key: NIFI-3411
>                 URL: https://issues.apache.org/jira/browse/NIFI-3411
>             Project: Apache NiFi
>          Issue Type: Improvement
>          Components: Core UI
>    Affects Versions: 1.1.1
>            Reporter: Joseph Gresock
>            Priority: Minor
>             Fix For: 1.2.0
>
>         Attachments: Status History 1.png, Status History 2.png
>
>
> In the Status History view on a NiFi cluster, the cluster's metrics (called "NiFi") are reported as a sum of the individual nodes' metrics, and appear as a blue line in the graph.  
> When the view is first displayed (or at any refresh), if any of the nodes have not reported back their metrics, the most recent data point is often misrepresented on the NiFi cluster line as the sum of only the nodes that have reported their metrics.  Since this is a line graph, the difference looks jarring, and can lead to the immediate interpretation that data has stopped flowing with such a large drop in the graph.  In addition, the summary metrics (avg/min/max) incorporate this incomplete cluster metric.  In order to get an accurate average data rate, the user has to continue to refresh the graph until all nodes have reported in the latest data point.
> I understand that this isn't a trivial fix (i.e., what do you do if one of the nodes genuinely is disconnected?), but it feels like this user experience could be improved.



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)