-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
https://reviews.apache.org/r/44213/
-----------------------------------------------------------
Review request for Ambari, Aravindan Vijayan, Dmytro Sen, and Sumit Mohanty.
Bugs: AMBARI-15250
https://issues.apache.org/jira/browse/AMBARI-15250
Repository: ambari
Description
-------
Found out that Ganglia added the processName in JVM metrics to make this
distinction::
org.apache.hadoop.metrics.ganglia.GangliaContext#emitRecord
{code}
// Setup so that the records have the proper leader names so they are
// unambiguous at the ganglia level, and this prevents a lot of rework
StringBuilder sb = new StringBuilder();
sb.append(contextName);
sb.append('.');
if (contextName.equals("jvm") && outRec.getTag("processName") != null) {
sb.append(outRec.getTag("processName"));
sb.append('.');
}
{code}
This is missing from our Sink which means some metrics do not have unique names
and will collide. _Affects HBase only_
Diffs
-----
ambari-metrics/ambari-metrics-hadoop-sink/src/main/java/org/apache/hadoop/metrics2/sink/timeline/HadoopTimelineMetricsSink.java
6da9257
Diff: https://reviews.apache.org/r/44213/diff/
Testing
-------
Manually verified metric names for JVM:
| jvm.Master.JvmMetrics.GcCount |
| jvm.Master.JvmMetrics.GcCountConcurrentMarkSweep |
| jvm.RegionServer.JvmMetrics.GcCountParNew |
| jvm.RegionServer.JvmMetrics.GcTimeMillis |
Thanks,
Sid Wagle
