You are viewing a plain text version of this content. The canonical link for it is here.
Posted to hdfs-user@hadoop.apache.org by Ivan Tretyakov <it...@griddynamics.com> on 2013/01/17 16:59:33 UTC

metrics all use the same namespace

Hi!

There is fixed issue in hadoop saying: "jvm metrics all use the same
namespace" - https://issues.apache.org/jira/browse/HADOOP-7507

I was able to apply this fix in our cluster using following line in
hadoop-metrics2.properties:
datanode.sink.ganglia.tagsForPrefix.jvm=*

So now I could see datanode jvm metrics with the following name, e.g.:
jvm.JvmMetrics.ProcessName_DataNode.GcCount

Also, it works for rpc namespace metrics, but looks little bit ugly because
of port number instead of service name:
rpc.rpc.port_50020.CallQueueLength

But it is not working for metricssystem and rpcdetails namespaces metrics
for example and may be other namespaces.
Configuration line we have:
datanode.sink.ganglia.tagsForPrefix.rpcdetailed=*
datanode.sink.ganglia.tagsForPrefix.metricssystem=*

The metrics name I could see , e.g.:
metricssystem.MetricsSystem.DroppedPubAll
rpcdetailed.rpcdetailed.InitReplicaRecoveryAvgTime

And from these names it is unclear which service sent it.
DataNode log shows following:

2013-01-17 15:01:36,706 DEBUG
org.apache.hadoop.metrics2.sink.ganglia.GangliaSink31: Emitting metric
metricssystem.MetricsSystem.Sink_gangliaAvgTime, type double, value 77.0,
slope both from hostname node07.dom.local
2013-01-17 15:01:36,706 DEBUG
org.apache.hadoop.metrics2.sink.ganglia.GangliaSink31: Emitting metric
metricssystem.MetricsSystem.NumAllSources, type int32, value 5, slope both
from hostname node07.dom.local
2013-01-17 15:01:36,706 DEBUG
org.apache.hadoop.metrics2.sink.ganglia.GangliaSink31: Emitting metric
metricssystem.MetricsSystem.PublishNumOps, type float, value 57, slope
positive from hostname node07.dom.local
2013-01-17 15:01:36,706 DEBUG
org.apache.hadoop.metrics2.sink.ganglia.GangliaSink31: Emitting metric
metricssystem.MetricsSystem.NumActiveSources, type int32, value 5, slope
both from hostname node07.dom.local

So, is it configuration issue, bug or something else?

Thanks in advance!

Hadoop version we use:
- MapReduce 2.0.0-mr1-cdh4.1.1
- HDFS 2.0.0-cdh4.1.1

-- 
Best Regards
Ivan Tretyakov