[
https://issues.apache.org/jira/browse/FLINK-8162?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16299240#comment-16299240
]
ASF GitHub Bot commented on FLINK-8162:
---------------------------------------
Github user tzulitai commented on a diff in the pull request:
https://github.com/apache/flink/pull/5182#discussion_r158161564
--- Diff:
flink-connectors/flink-connector-kinesis/src/main/java/org/apache/flink/streaming/connectors/kinesis/internals/ShardConsumer.java
---
@@ -96,6 +99,15 @@ protected ShardConsumer(KinesisDataFetcher<T> fetcherRef,
SequenceNumber
lastSequenceNum,
KinesisProxyInterface
kinesis) {
this.fetcherRef = checkNotNull(fetcherRef);
+ MetricGroup kinesisMetricGroup = fetcherRef.getRuntimeContext()
--- End diff --
I feel that passing the `StreamingRuntimeContext` all the way here just to
register metrics, is not a good idea.
Is it possible we register the metrics in `FlinkKinesisConsumer` instead?
That also makes it more visible what metrics the consumer exposes without
having to dig all the way to this internal `ShardConsumer` thread.
> Kinesis Connector to report millisBehindLatest metric
> -----------------------------------------------------
>
> Key: FLINK-8162
> URL: https://issues.apache.org/jira/browse/FLINK-8162
> Project: Flink
> Issue Type: Improvement
> Components: Kinesis Connector
> Reporter: Cristian
> Priority: Minor
> Labels: kinesis
> Fix For: 1.5.0
>
> Original Estimate: 24h
> Remaining Estimate: 24h
>
> When reading from Kinesis streams, one of the most valuable metrics is
> "MillisBehindLatest" (see
> https://github.com/aws/aws-sdk-java/blob/25f0821f69bf94ec456f602f2b83ea2b0ca15643/aws-java-sdk-kinesis/src/main/java/com/amazonaws/services/kinesis/model/GetRecordsResult.java#L187-L201).
> Flink should use its metrics mechanism to report this value as a gauge,
> tagging it with the shard id.
--
This message was sent by Atlassian JIRA
(v6.4.14#64029)
