-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
https://reviews.apache.org/r/50353/#review143263
-----------------------------------------------------------


Fix it, then Ship it!





ambari-metrics/ambari-metrics-timelineservice/src/test/java/org/apache/hadoop/yarn/server/applicationhistoryservice/metrics/timeline/aggregators/TimelineMetricClusterAggregatorSecondTest.java
 (line 123)
<https://reviews.apache.org/r/50353/#comment209042>

    Logger in stead of System.out



ambari-metrics/ambari-metrics-timelineservice/src/test/java/org/apache/hadoop/yarn/server/applicationhistoryservice/metrics/timeline/aggregators/TimelineMetricClusterAggregatorSecondTest.java
 (line 138)
<https://reviews.apache.org/r/50353/#comment209043>

    Like this comment.



ambari-metrics/ambari-metrics-timelineservice/src/test/java/org/apache/hadoop/yarn/server/applicationhistoryservice/metrics/timeline/aggregators/TimelineMetricClusterAggregatorSecondTest.java
 (line 243)
<https://reviews.apache.org/r/50353/#comment209044>

    Comment issue



ambari-metrics/ambari-metrics-timelineservice/src/test/java/org/apache/hadoop/yarn/server/applicationhistoryservice/metrics/timeline/aggregators/TimelineMetricClusterAggregatorSecondTest.java
 (line 280)
<https://reviews.apache.org/r/50353/#comment209045>

    Ideally this should be govered by a policy in the configuration. Maybe we 
should handle this as a separate 3.0 Jira.


- Sid Wagle


On July 22, 2016, 8:08 p.m., Aravindan Vijayan wrote:
> 
> -----------------------------------------------------------
> This is an automatically generated e-mail. To reply, visit:
> https://reviews.apache.org/r/50353/
> -----------------------------------------------------------
> 
> (Updated July 22, 2016, 8:08 p.m.)
> 
> 
> Review request for Ambari, Dmytro Sen, Sumit Mohanty, and Sid Wagle.
> 
> 
> Bugs: AMBARI-17845
>     https://issues.apache.org/jira/browse/AMBARI-17845
> 
> 
> Repository: ambari
> 
> 
> Description
> -------
> 
> PROBLEM
> Storm cluster metrics (aggregated across hosts) don't show up in Ambari and 
> Grafana
> 
> BUG
> Storm metrics collect and send data in 1 minute intervals. Since their data 
> is present in the right end of the spectrum for the 2 minute aggregator 
> (start_time ~ server_time), a bug in the second aggregator is causing these 
> values to slip between 2 aggregator cycles.
> 
> FIX
> Query the database for 3 minute interval, look for the data in the time 
> shifted interval ( 
> ams-site:timeline.metrics.service.cluster.aggregator.timeshift.adjustment). 
> In case no data is present, look for the data outside the right boundary of 
> the interval. Use that to interpolate the data in the 30second slices.
> 
> 
> Diffs
> -----
> 
>   
> ambari-metrics/ambari-metrics-timelineservice/src/main/java/org/apache/hadoop/yarn/server/applicationhistoryservice/metrics/timeline/aggregators/TimelineMetricClusterAggregatorSecond.java
>  bd46045 
>   
> ambari-metrics/ambari-metrics-timelineservice/src/test/java/org/apache/hadoop/yarn/server/applicationhistoryservice/metrics/timeline/aggregators/TimelineMetricClusterAggregatorSecondTest.java
>  6776a3c 
> 
> Diff: https://reviews.apache.org/r/50353/diff/
> 
> 
> Testing
> -------
> 
> Manually tested.
> Added detailed unit test.
> mvn clean test
> 
> 
> Thanks,
> 
> Aravindan Vijayan
> 
>

Reply via email to