Packt+ | Advance your knowledge in tech

You're reading from Monitoring Hadoop

Product typeBook

Published inApr 2015

Publisher

ISBN-139781783281558

Edition1st Edition

Tools

Hadoop

Concepts

Data Processing

Author (1)

Aman Singh

Chapter 6. Hadoop Metrics and Visualization Using Ganglia

In this chapter, we will look at the Hadoop metrics and visualization of various components like CPU, memory, and disk, by using Ganglia. This chapter is a build from the initial chapters on the monitoring and installation of Ganglia. Hadoop is a distributed platform with various services running across the cluster, which provides many metrics to tap into the Hadoop counters and other functional parameters.

In this chapter, we will look at the metrics for various Hadoop components.

The following topics will be covered in this chapter:

Hadoop metrics contexts
Metrics collection under DFS context
Metrics collection under mapred context
Metrics collection under RPC, JVM, and other contexts
Visualizing the metrics with Ganglia

Hadoop metrics

In Hadoop, there are many daemons running, such as DataNode, NameNode, and JobTracker; each of these daemons captures a lot of information about the components they work on. Similarly, in the YARN framework, we have ResourceManager, NodeManager, and ApplicationManager, each of which exposes metrics, explained in the following sections under Metrics2. For example, DataNode collects metrics such as the number of blocks it has for advertising to the NameNode, the number of replicated blocks, and metrics about read/writes from clients. In addition to this, there could be metrics related to events, and so on. Hence, it is very important to gather them for the working of the Hadoop cluster and for debugging, if something goes wrong.

Therefore, Hadoop has a metrics system for collecting all this information. There are two versions of the metrics system, metrics and Metrics2 for Hadoop 1.x and Hadoop 2.x, respectively. The hadoop-metrics.properties and hadoop-metrics2.properties files...

Metrics contexts

Metrics are more relevant to the maintainers of the Hadoop clusters than its users. There might be many users who run MapReduce jobs on a cluster; they are concerned about MapReduce Counters and not the metrics, which are daemon specific. MapReduce counters talk about the number of mappers or reducers, number of bytes read or written to the HDFS and non-HDFS File System, how many spills happened, information about the shuffle phase, etc. However, for Hadoop administrators, metrics about the daemons are of more concern, in order to better understand the cluster.

Named contexts

Each of the daemons has a group of contexts for it. Some of the contexts, which are supported or rather available, are listed in the following table:

Metrics system design

Hadoop provides a framework to collect internal events and metrics and report them to the external system. The external system could be simply writing to a file or a tool like Ganglia. The new Hadoop Metrics2 framework has been revamped to integrate better with Ganglia.

The best things about the framework are the pluggable output plugins and the ability to reconfigure it without the need to restart the daemons.

The metrics have three main parts:

Producer: The producer is the source of metrics generation and produces metrics for the upstream
Consumers: They are basically the sinks of the framework, as they consume the metrics generated by the producers
Pollers: They poll the sources and deliver data to the sink or consumers

Metrics configuration

The Hadoop daemons expose runtime metrics, which can be collected using plugins. The old Metrics1 system has been replaced by the new Metrics2 system, which supports the following:

Metrics collection using multiple plugins
Better integration with JMX
Better filters for cutting out noise

Before configuring metrics, it is important to understand which metrics and servlets are supported by each Hadoop version. For example, the servlet at /metrics works only with Metrics1 and the new servlet at /jmx works with both Metrics1 and Metrics2.

We need to configure a source, consumer, and poller for the framework:

Source or producer: A metric source class must implement the following interface:
```
org.apache.hadoop.metrics2.MetricsSource
```
Consumer or sink: A consumer or sink must be implemented with the following line of code:
```
org.apache.hadoop.metrics2.MetricsSink
```
For example, the configuration for JobTracker sink and filter is as follows:
```
jobtracker.sink.file.class=org.apache.hadoop.metrics2...
```

Configuring Metrics2

For Hadoop version 2, which uses the YARN framework, the metrics can be configured using hadoop-metrics2.properties, in the $HADOOP_HOME folder:

*.sink.file.class=org.apache.hadoop.metrics2.sink.FileSink
*.period=10
namenode.sink.file.filename=namenode-metrics.out
datanode.sink.file.filename=datanode-metrics.out
jobtracker.sink.file.filename=jobtracker-metrics.out
tasktracker.sink.file.filename=tasktracker-metrics.out
maptask.sink.file.filename=maptask-metrics.out
reducetask.sink.file.filename=reducetask-metrics.out

We can also script it out and use it for metrics generation, shown as follows:

# namenode
[script://./bin/hadoop_metrics.sh http://192.168.1.70:50070/jmx]
disabled = 0
interval = 10
sourcetype = hadoop_metrics
index = hadoop_metrics

# datanode
[script://./bin/hadoop_metrics.sh http://192.168.1.70:50075/jmx]
disabled = 0
interval = 10
sourcetype = hadoop_metrics
index = hadoop_metrics

# jobtracker
[script://./bin/hadoop_metrics.sh http://192.168.1.70:50030...

Exploring the metrics contexts

Till now, we have seen that there are various metrics contexts such as JVM, DFS, and RPC. Let's look at them and explore some of the examples, depicting what each context looks like and what it logs:

JVM context: The JVM context contains stats about JVM memory, threads, heap memory, and so on:

jvm.metrics: hostName=dn1.cluster1.com, processName=DataNode, sessionId=,logError=0,logFatal=0,logInfo=159,logWarn=0, memHeapCommittedM=9.4,memHeapUsedM=12.63,memNonHeapCommittedM=28.75,memNonHeapUsedM=19.7356,threadsBlocked=0, threadsNew=0, threadsRunnable=3, threadsTerminated=0, threadsTimedWaiting=2, threadsWaiting=1

DFS context: The DFS context stats talk about the namenode files, capacity, blocks, and so on:

dfs.FSNamesystem: hostName=nn1.cluster1.com, sessionId=, BlocksTotal=440, CapacityRemainingGB=100, CapacityTotalGB=254, CapacityUsedGB=0, FilesTotal=160, PendingReplicationBlocks=0, ScheduledReplicationBlocks=0, TotalLoad=1, UnderReplicatedBlocks=20

Mapred context...

Hadoop Ganglia integration

Ganglia is a metrics collection and a visualization tool for the enterprise and works very well with Nagios and Hadoop. In addition to just collecting stats about CPU, memory, and disk, other finely tuned metrics are required, which can be provided by this framework.

Until now, we have seen that the metrics collection can be done to a file or to any other tool like Splunk, depending upon the class interface. We can configure which class handles the metrics update.

For Ganglia, we use GangliaContext, which is an implementation of MetricsContext. Ganglia versions higher than 3.0 provide this integration and work very well for collecting the Hadoop metrics.

In Ganglia, the metrics can be collected for NameNode, JobTracker, MapReduce tasks, JVM, RPC, DataNodes, and the new YARN framework.

Hadoop metrics configuration for Ganglia

Firstly, we need to define a sink class, as per Ganglia version 3.1:

*.sink.ganglia.class=org.apache.hadoop.metrics2.sink.ganglia.GangliaSink31

Secondly...

Hadoop configuration

Now, we must set up the Hadoop Configuration file to point to the Ganglia servers.

Metrics1

Update the hadoop-metrics.properties file with the following lines:

dfs.class=org.apache.hadoop.metrics.ganglia.GangliaContext31
dfs.period=10
dfs.servers=192.168.1.10:8649

mapred.class=org.apache.hadoop.metrics.ganglia.GangliaContext31
mapred.period=10
mapred.servers=192.168.1.10:8649

jvm.class=org.apache.hadoop.metrics.ganglia.GangliaContext31
jvm.period=10
jvm.servers=192.168.1.10:8649

rpc.class=org.apache.hadoop.metrics.ganglia.GangliaContext31
rpc.period=10
rpc.servers=192.168.1.10:8649

ugi.class=org.apache.hadoop.metrics.ganglia.GangliaContext31
ugi.period=10
ugi.servers=192.168.1.10:8649

Metrics2

Update the hadoop-metrics2.properties file with the following lines:

namenode.sink.ganglia.class=org.apache.hadoop.metrics2.sink.ganglia.GangliaSink31
namenode.sink.ganglia.period=30
namenode.sink.ganglia.servers=192.168.1.10:8649

datanode.sink.ganglia.class=org.apache.hadoop.metrics2...

Ganglia graphs

Once the configuration is in place and the services have started, we can see the metrics being collected and plotted by using the Ganglia web interface, as shown in the following screenshot:

Let's have a look at the next screenshot:

Metrics APIs

For reporting metrics, we have a package that provides APIs for both Metrics1 and Metrics2. It provides the flexibility to use client libraries and different modules from within an application.

The org.apache.hadoop.metrics package

This package provides sub-packages to do the specified task:

org.apache.hadoop.metrics.spi

The abstract Server Provider Interface package. Those wishing to integrate the metrics API with a particular metrics client library should extend this package:

org.apache.hadoop.metrics.file

An implementation package that writes the metric data to a file or sends it to the standard output stream:

org.apache.hadoop.metrics.ganglia

An implementation package that sends metrics data to Ganglia.

The new Metrics2 provides a lot more packages for the implementation.

The org.apache.hadoop.metrics2 package

org.apache.hadoop.metrics2.annotation: This is the public annotation that interfaces for simpler metrics instrumentation.
org.apache.hadoop.metrics2.filter: This is the built...

Summary

In this chapter, we looked at how to do metrics collections, the different metrics contexts and their groups, and the package APIs for integration with Ganglia for graphing the metrics. In the next chapter, we will look at the monitoring of some of the other components of Hadoop, such as Hive and HBase, and some performance improvement tips and tuning.

The rest of the chapter is locked

You have been reading a chapter from

Monitoring Hadoop

Published in: Apr 2015Publisher: ISBN-13: 9781783281558

A free Packt account unlocks extra newsletters, articles, discounted offers, and much more. Start advancing your knowledge today.

undefined

Unlock this book and the full library FREE for 7 days

Get unlimited access to 7000+ expert-authored eBooks and videos courses covering every tech area you can think of

Start free trial

Renews at $15.99/month. Cancel anytime

Author (1)

Aman Singh

Gurmukh Singh is a seasoned technology professional with 14+ years of industry experience in infrastructure design, distributed systems, performance optimization, and networks. He has worked in big data domain for the last 5 years and provides consultancy and training on various technologies. He has worked with companies such as HP, JP Morgan, and Yahoo. He has authored Monitoring Hadoop by Packt Publishing
Read more about Aman Singh

Personalised recommendations for you

Based on your interests and search pattern

Et al.

Ever wonder why speech recognition systems don't understand the Scottish accent, or what would happen if an astronaut only ate mac 'n' cheese, or other spurious reflections you'd have at a bar? We did, then collated those deliberations into absurd research articles with fake figures and methodologies inspired by even more fictionally absurd studies.

BookAug 2023230 pages5

Generative AI with LangChain

This book is a comprehensive introduction to LLMs and LangChain, demystifying the basic mechanics of LangChain, its functionalities, and the myriad of applications it can be integrated into.

BookDec 2023360 pages4

Generative AI with LangChain

This book is a comprehensive introduction to LLMs and LangChain, demystifying the basic mechanics of LangChain, its functionalities, and the myriad of applications it can be integrated into.

BookDec 2023360 pages5

Generative AI with LangChain

This book is a comprehensive introduction to LLMs and LangChain, demystifying the basic mechanics of LangChain, its functionalities, and the myriad of applications it can be integrated into.

BookDec 2023360 pages1

Generative AI with LangChain

This book is a comprehensive introduction to LLMs and LangChain, demystifying the basic mechanics of LangChain, its functionalities, and the myriad of applications it can be integrated into.

BookDec 2023360 pages5

Mastering Tableau 2023

This book is a comprehensive resource to mastering your Tableau skills and becoming a BI expert. As you progress, you will learn how to build advanced dashboards and improve your storytelling to derive key business insight, as well as make you well-versed with advanced functionalities of Tableau in the business intelligence domain.

BookAug 2023684 pages

Building AI Applications with ChatGPT APIs

This guide covers all ChatGPT API features for effortless creation of robust AI powered apps. With its help, you’ll be able to leverage ChatGPT’s cutting-edge NLP models to take your app development skills to the next level. You’ll also work on ten exciting projects that will give you the practical know-how that you can apply to your existing applications.

BookSep 2023258 pages5

Building AI Applications with ChatGPT APIs

This guide covers all ChatGPT API features for effortless creation of robust AI powered apps. With its help, you’ll be able to leverage ChatGPT’s cutting-edge NLP models to take your app development skills to the next level. You’ll also work on ten exciting projects that will give you the practical know-how that you can apply to your existing applications.

BookSep 2023258 pages2

Data Engineering with AWS

Embark on a journey to master data engineering pipelines on AWS! Our book offers a hands-on experience of AWS services for ingesting, transforming, and consuming data. Whether you're an absolute beginner or someone with basic data engineering experience, this guide is an indispensable resource.

BookOct 2023636 pages5

Modern Data Architecture on AWS

Every organization wants an agile, performant, and cost-effective data platform that meets all their current and future business needs. Purpose-built AWS analytics services and their features play a big part in building such a modern data platform. This book brings to you all the design and architectural patterns that’ll help you achieve this goal.

BookAug 2023420 pages5

Practical Guide to Applied Conformal Prediction in Python

Discover the power of Conformal Prediction with the "Practical Guide to Applied Conformal Prediction in Python." Master the latest techniques to quantify uncertainty in machine learning and computer vision models, and seamlessly apply them to your industry applications.

BookDec 2023240 pages

TinyML Cookbook

With over 70 project-based recipes, the TinyML Cookbook is a practical guide that will help you to get the most out of your microcontrollers. It provides a comprehensive understanding of the theoretical foundations while giving you hands-on experience training ML models for deployment on Arduino Nano 33 BLE Sense, Raspberry Pi Pico, and SparkFun RedBoard Artemis Nano microcontrollers.

BookNov 2023664 pages

Hadoop 1.x	Hadoop 2.x
jvm: for Java Virtual Machine	yarn: for the YARN components
dfs: for Distributed File System	jvm: for Java Virtual Machine
mapred: for JobTracker and TaskTracker	dfs: for Distributed File System
rpc: for Remote Procedure...