You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@flink.apache.org by "Chesnay Schepler (JIRA)" <ji...@apache.org> on 2019/04/26 10:15:06 UTC

[jira] [Commented] (FLINK-12103) Extend builtin operator metrics with user defined scope and variables

    [ https://issues.apache.org/jira/browse/FLINK-12103?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16826853#comment-16826853 ] 

Chesnay Schepler commented on FLINK-12103:
------------------------------------------

Can you provide more details on what kind of variables you'd want do add? Your example effectively implemented via the task_name variable.

> Extend builtin operator metrics with user defined scope and variables
> ---------------------------------------------------------------------
>
>                 Key: FLINK-12103
>                 URL: https://issues.apache.org/jira/browse/FLINK-12103
>             Project: Flink
>          Issue Type: Wish
>          Components: Runtime / Metrics
>    Affects Versions: 1.7.2
>            Reporter: Ben Marini
>            Priority: Minor
>
> I'm looking for a way to extend the builtin throughput and latency metrics for operators with my own metric variables.
> My specific use case:
> I have a job that defines a list of independent source -> sink streams. I would like to add my own metric variables to each of these independent streams. For example, something like this:
> {code:scala}
> class MyFilter extends RichFilterFunction {
>   override def open(parameters: Configuration): Unit = {
>     val mg = getRuntimeContext.getMetricGroup // Includes "streamName" -> "A|B"
>     // Init some user defined metrics here...
>   }
> }
> // Stream A
> // Native operator metrics and user defined metrics in rich operators include "streamName" -> "A"
> streamA = env.withMetricGroup((mg) => mg.addGroup("streamName", "A").addSource(...).filter(new MyFilter).addSink(...)
> // Stream B
> // Native operator metrics and user defined metrics in rich operators include "streamName" -> "B"
> streamB = env.withMetricGroup((mg) => mg.addGroup("streamName", "B").addSource(...).filter(new MyFilter).addSink(...)
> {code}
>  
> Is this possible? Would a new hook into StreamTransformation have to be added?



--
This message was sent by Atlassian JIRA
(v7.6.3#76005)