[
https://issues.apache.org/jira/browse/BEAM-7438?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17143747#comment-17143747
]
Maximilian Michels commented on BEAM-7438:
------------------------------------------
Sorry for the confusion, I was talking about the portable metrics results in
Python where Gauges and Distributions are supported. It is true that the Flink
metric reporting still does not work correctly.
Looks good to me to map Gauge to Long and Distribution to Histogram. Once Beam
has a histogram, we can also add support via Flink's histogram.
> Distribution and Gauge metrics are not being exported to Flink dashboard
> neither Prometheus IO
> ----------------------------------------------------------------------------------------------
>
> Key: BEAM-7438
> URL: https://issues.apache.org/jira/browse/BEAM-7438
> Project: Beam
> Issue Type: Bug
> Components: runner-flink
> Affects Versions: 2.12.0, 2.13.0
> Reporter: Ricardo Bordon
> Priority: P2
> Attachments: image-2019-05-29-11-24-36-911.png,
> image-2019-05-29-11-26-49-685.png
>
>
> Distributions and gauge metrics are not visible at Flink dashboard neither
> Prometheus IO.
> I was able to debug the runner code and see that these metrics are being
> update over *FlinkMetricContainer#updateDistributions()* andÂ
> *FlinkMetricContainer#updateGauges()* (meaning they are treated as "attempted
> metrics") but these are not visible when looking them over the Flink
> Dashboard or Prometheus. In the other hand, *Counter* metrics work as
> expected.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
