[
https://issues.apache.org/jira/browse/FLINK-5069?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15709675#comment-15709675
]
ASF GitHub Bot commented on FLINK-5069:
---------------------------------------
Github user zentol commented on the issue:
https://github.com/apache/flink/pull/2809
I don't think There is a benefit in having a separate custom class (in this
case, PendingCheckpointStats) to encapsulate the data.
Most reporters turn Gauge values to strings. This means that the list and
the PendingCheckpoitnStats objects are pretty much immediately discarded. We
can save some overhead by creating the String straight away.
The only reporter that doesn't turn them to strings is the JMXReporter,
which however would not be able to expose this object at all since it doesn't
fulfill the necessary requirements.
> Pending checkpoint statistics gauge
> -----------------------------------
>
> Key: FLINK-5069
> URL: https://issues.apache.org/jira/browse/FLINK-5069
> Project: Flink
> Issue Type: Improvement
> Components: Metrics
> Reporter: zhuhaifeng
> Assignee: zhuhaifeng
> Priority: Minor
> Fix For: 1.2.0
>
>
> Add the statistics of pending checkpoints as a gauge metric. When the
> checkpoint appears not to be completed, this metric would help to get the
> state of a pending checkpoint , e.g which task did not complete the
> checkpoint.
> The statistic will be as the follows：
> checkpointID,
> Number Of Acknowledged Tasks,
> Number Of Not yet Acknowledged Tasks,
> The detail of not yet acknowledged JobVertexID, taskID
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
