Hadoop On Demand
Introduction
The Hadoop On Demand (HOD) project is a system for provisioning and managing independent Hadoop MapReduce instances on a shared cluster of nodes. HOD uses a resource manager for allocation. At present it supports Torque out of the box.
Feature List
Simplified Interface for Provisioning Hadoop Clusters
By far, the biggest advantage of HOD is to quickly setup a Hadoop cluster. The user interacts with the cluster through a simple command line interface, the HOD client. HOD brings up a virtual MapReduce cluster with the required number of nodes, which the user can use for running Hadoop jobs. When done, HOD will automatically clean up the resources and make the nodes available again.
Automatic installation of Hadoop
With HOD, Hadoop does not need to be even installed on the cluster. The user can provide a Hadoop tarball that HOD will automatically distribute to all the nodes in the cluster.
Configuring Hadoop
Dynamic parameters of Hadoop configuration, such as the NameNode and JobTracker addresses and ports, and file system temporary directories are generated and distributed by HOD automatically to all nodes in the cluster. In addition, HOD allows the user to configure Hadoop parameters at both the server (for e.g. JobTracker) and client (for e.g. JobClient) level, including 'final' parameters, that were introduced with Hadoop 0.15.
Auto-cleanup of Unused Clusters
HOD has an automatic timeout so that users cannot misuse resources they aren't using. The timeout applies only when there is no MapReduce job running.
Log Services
HOD can be used to collect all MapReduce logs to a central location for archiving and inspection after the job is completed.
HOD Components
This is a brief overview of the various components of HOD and how they interact to provision Hadoop.
HOD Client
The HOD client is a Unix command that users use to allocate Hadoop MapReduce clusters. The command provides other options to list allocated clusters and deallocate them. The HOD client generates the hadoop-site.xml in a user specified directory. The user can point to this configuration file while running Map/Reduce jobs on the allocated cluster.
The nodes from where the HOD Client is run are called submit nodes because jobs are submitted to the resource manager system for allocating and running clusters from these nodes.
RingMaster
The RingMaster is a HOD process that is started on one node per every allocated cluster. It is submitted as a 'job' to the resource manager by the HOD client. It controls which Hadoop daemons start on which nodes. It provides this information to other HOD processes, such as the HOD client, so users can also determine this information. The RingMaster is responsible for hosting and distributing the Hadoop tarball to all nodes in the cluster. It also automatically cleans up unused clusters.
HodRing
The HodRing is a HOD process that runs on every allocated node in the cluster. These processes are run by the RingMaster through the resource manager, using a facility of parallel execution. The HodRings are responsible for launching Hadoop commands on the nodes to bring up the Hadoop daemons. They get the command to launch from the RingMaster.
Hodrc / HOD configuration file
An INI style configuration file where the users configure various options for the HOD system, including install locations of different software, resource manager parameters, log and temp file directories, parameters for their MapReduce jobs, etc.
Submit Nodes and Compute Nodes
The nodes from where the HOD Client is run are referred as submit nodes because jobs are submitted to the resource manager system for allocating and running clusters from these nodes.
The nodes where the Ringmaster and HodRings run are called the Compute nodes. These are the nodes that get allocated by a resource manager, and on which the Hadoop daemons are provisioned and started.
Getting Started with HOD
Pre-Requisites
Hardware
HOD requires a minimum of 3 nodes configured through a resource manager.
Software
The following components are assumed to be installed before using HOD:
- Torque: Currently HOD supports Torque out of the box. We assume that you are familiar with configuring Torque. You can get information about this from here.
- Python: We require version 2.5.1, which can be downloaded from here.
The following components can be optionally installed for getting better functionality from HOD:
- Twisted Python: This can be used for improving the scalability of HOD. Twisted Python is available here.
- Hadoop: HOD can automatically distribute Hadoop to all nodes in the cluster. However, it can also use a pre-installed version of Hadoop, if it is available on all nodes in the cluster. HOD currently supports Hadoop 0.15 and above.
HOD configuration requires the location of installs of these components to be the same on all nodes in the cluster. It will also make the configuration simpler to have the same location on the submit nodes.
Resource Manager Configuration Pre-requisites
For using HOD with Torque:
- Install Torque components: pbs_server on a head node, pbs_moms on all compute nodes, and PBS client tools on all compute nodes and submit nodes.
- Create a queue for submitting jobs on the pbs_server.
- Specify a name for all nodes in the cluster, by setting a 'node property' to all the nodes. This can be done by using the 'qmgr' command. For example: qmgr -c "set node node properties=cluster-name"
- Ensure that jobs can be submitted to the nodes. This can be done by using the 'qsub' command. For example: echo "sleep 30" | qsub -l nodes=3
More information about setting up Torque can be found by referring to the documentation here.
Setting up HOD
- HOD is available in the 'contrib' section of Hadoop under the root directory 'hod'. Distribute the files under this directory to all the nodes in the cluster.
- On the node from where you want to run hod, edit the file hodrc which can be found in the install dir/conf directory. This file contains the minimal set of values required for running hod.
- Specify values suitable to your environment for the following variables defined in the configuration file. Note that some of these variables are defined at more than one place in the file.
| Variable Name | Meaning |
|---|---|
| ${JAVA_HOME} | Location of Java for Hadoop. Hadoop supports Sun JDK 1.5.x |
| ${CLUSTER_NAME} | Name of the cluster which is specified in the 'node property' as mentioned in resource manager configuration. |
| ${HADOOP_HOME} | Location of Hadoop installation on the compute and submit nodes. |
| ${RM_QUEUE} | Queue configured for submiting jobs in the resource manager configuration. |
| ${RM_HOME} | Location of the resource manager installation on the compute and submit nodes. |
- The following environment variables *may* need to be set depending on your environment. These variables must be defined where you run the HOD client, and also be specified in the HOD configuration file as the value of the key resource_manager.env-vars. Multiple variables can be specified as a comma separated list of key=value pairs.
| Variable Name | Meaning |
|---|---|
| HOD_PYTHON_HOME | If you install python to a non-default location of the compute nodes, or submit nodes, then, this variable must be defined to point to the python executable in the non-standard location. |
You can also review other configuration options in the file and modify them to suit your needs. Refer to the the section on configuration below for information about the HOD configuration.
Running HOD
Overview
A typical session of HOD will involve atleast three steps: allocate, run hadoop jobs, deallocate.
Operation allocate
The allocate operation is used to allocate a set of nodes and install and provision Hadoop on them. It has the following syntax:
| hod -c config_file -t hadoop_tarball_location -o "allocate cluster_dir number_of_nodes" |
The hadoop_tarball_location must be a location on a shared file system accesible from all nodes in the cluster. Note, the cluster_dir must exist before running the command. If the command completes successfully then cluster_dir/hadoop-site.xml will be generated and will contain information about the allocated cluster's JobTracker and NameNode.
For example, the following command uses a hodrc file in ~/hod-config/hodrc and allocates Hadoop (provided by the tarball ~/share/hadoop.tar.gz) on 10 nodes, storing the generated Hadoop configuration in a directory named ~/hadoop-cluster:
| $ hod -c ~/hod-config/hodrc -t ~/share/hadoop.tar.gz -o "allocate ~/hadoop-cluster 10" |
HOD also supports an environment variable called HOD_CONF_DIR. If this is defined, HOD will look for a default hodrc file at $HOD_CONF_DIR/hodrc. Defining this allows the above command to also be run as follows:
|
$ export HOD_CONF_DIR=~/hod-config $ hod -t ~/share/hadoop.tar.gz -o "allocate ~/hadoop-cluster 10" |
Running Hadoop jobs using the allocated cluster
Now, one can run Hadoop jobs using the allocated cluster in the usual manner:
| hadoop --config cluster_dir hadoop_command hadoop_command_args |
Continuing our example, the following command will run a wordcount example on the allocated cluster:
| $ hadoop --config ~/hadoop-cluster jar /path/to/hadoop/hadoop-examples.jar wordcount /path/to/input /path/to/output |
Operation deallocate
The deallocate operation is used to release an allocated cluster. When finished with a cluster, deallocate must be run so that the nodes become free for others to use. The deallocate operation has the following syntax:
| hod -o "deallocate cluster_dir" |
Continuing our example, the following command will deallocate the cluster:
| $ hod -o "deallocate ~/hadoop-cluster" |
Command Line Options
This section covers the major command line options available via the hod command:
--help
Prints out the help message to see the basic options.
--verbose-help
All configuration options provided in the hodrc file can be passed on the command line, using the syntax --section_name.option_name[=value]. When provided this way, the value provided on command line overrides the option provided in hodrc. The verbose-help command lists all the available options in the hodrc file. This is also a nice way to see the meaning of the configuration options.
-c config_file
Provides the configuration file to use. Can be used with all other options of HOD. Alternatively, the HOD_CONF_DIR environment variable can be defined to specify a directory that contains a file named hodrc, alleviating the need to specify the configuration file in each HOD command.
-b 1|2|3|4
Enables the given debug level. Can be used with all other options of HOD. 4 is most verbose.
-o "help"
Lists the operations available in the operation mode.
-o "allocate cluster_dir number_of_nodes"
Allocates a cluster on the given number of cluster nodes, and store the allocation information in cluster_dir for use with subsequent hadoop commands. Note that the cluster_dir must exist before running the command.
-o "list"
Lists the clusters allocated by this user. Information provided includes the Torque job id corresponding to the cluster, the cluster directory where the allocation information is stored, and whether the Map/Reduce daemon is still active or not.
-o "info cluster_dir"
Lists information about the cluster whose allocation information is stored in the specified cluster directory.
-o "deallocate cluster_dir"
Deallocates the cluster whose allocation information is stored in the specified cluster directory.
-t hadoop_tarball
Provisions Hadoop from the given tar.gz file. This option is only applicable to the allocate operation. For better distribution performance it is recommended that the Hadoop tarball contain only the libraries and binaries, and not the source or documentation.
-Mkey1=value1 -Mkey2=value2
Provides configuration parameters for the provisioned Map/Reduce daemons (JobTracker and TaskTrackers). A hadoop-site.xml is generated with these values on the cluster nodes
-Hkey1=value1 -Hkey2=value2
Provides configuration parameters for the provisioned HDFS daemons (NameNode and DataNodes). A hadoop-site.xml is generated with these values on the cluster nodes
-Ckey1=value1 -Ckey2=value2
Provides configuration parameters for the client from where jobs can be submitted. A hadoop-site.xml is generated with these values on the submit node.
HOD Configuration
Introduction to HOD Configuration
Configuration options for HOD are organized as sections and options within them. They can be specified in two ways: a configuration file in the INI format, and as command line options to the HOD shell, specified in the format --section.option[=value]. If the same option is specified in both places, the value specified on the command line overrides the value in the configuration file.
To get a simple description of all configuration options, you can type hod --verbose-help
This section explains some of the most important or commonly used configuration options in some more detail.
Categories / Sections in HOD Configuration
The following are the various sections in the HOD configuration:
| Section Name | Description |
|---|---|
| hod | Options for the HOD client |
| resource_manager | Options for specifying which resource manager to use, and other parameters for using that resource manager |
| ringmaster | Options for the RingMaster process |
| hodring | Options for the HodRing process |
| gridservice-mapred | Options for the MapReduce daemons |
| gridservice-hdfs | Options for the HDFS daemons |
Important and Commonly Used Configuration Options
Common configuration options
Certain configuration options are defined in most of the sections of the HOD configuration. Options defined in a section, are used by the process for which that section applies. These options have the same meaning, but can have different values in each section.
| Option Name | Description |
|---|---|
| temp-dir | Temporary directory for usage by the HOD processes. Make sure that the users who will run hod have rights to create directories under the directory specified here. |
| debug | A numeric value from 1-4. 4 produces the most log information, and 1 the least. |
| log-dir | Directory where log files are stored. By default, this is install-location/logs/. The restrictions and notes for the temp-dir variable apply here too. |
| xrs-port-range | A range of ports, among which an available port shall be picked for use to run any XML-RPC based server daemon processes of HOD. |
| http-port-range | A range of ports, among which an available port shall be picked for use to run any HTTP based server daemon processes of HOD. |
hod options
| Option Name | Description |
|---|---|
| cluster | A descriptive name given to the cluster. For Torque, this is specified as a 'Node property' for every node in the cluster. HOD uses this value to compute the number of available nodes. |
| client-params | A comma-separated list of hadoop config parameters specified as key-value pairs. These will be used to generate a hadoop-site.xml on the submit node that should be used for running MapReduce jobs. |
resource_manager options
| Option Name | Description |
|---|---|
| queue | Name of the queue configured in the resource manager to which jobs are to be submitted. |
| batch-home | Install directory to which 'bin' is appended and under which the executables of the resource manager can be found. |
| env-vars | This is a comma separated list of key-value pairs, expressed as key=value, which would be passed to the jobs launched on the compute nodes. For example, if the python installation is in a non-standard location, one can set the environment variable 'HOD_PYTHON_HOME' to the path to the python executable. The HOD processes launched on the compute nodes can then use this variable. |
ringmaster options
| Option Name | Description |
|---|---|
| work-dirs | These are a list of comma separated paths that will serve as the root for directories that HOD generates and passes to Hadoop for use to store DFS / MapReduce data. For e.g. this is where DFS data blocks will be stored. Typically, as many paths are specified as there are disks available to ensure all disks are being utilized. The restrictions and notes for the temp-dir variable apply here too. |
gridservice-hdfs options
| Option Name | Description |
|---|---|
| external |
If false, this indicates that a HDFS cluster must be bought up by the HOD system, on the nodes which it allocates via the allocate command. Note that in that case, when the cluster is de-allocated, it will bring down the HDFS cluster, and all the data will be lost. If true, it will try and connect to an externally configured HDFS system. Typically, because input for jobs are placed into HDFS before jobs are run, and also the output from jobs in HDFS is required to be persistent, an internal HDFS cluster is of little value in a production system. However, it allows for quick testing. |
| host | Hostname of the externally configured NameNode, if any. |
| fs_port | Port to which NameNode RPC server is bound. |
| info_port | Port to which the NameNode web UI server is bound. |
| pkgs | Installation directory, under which bin/hadoop executable is located. This can be used to use a pre-installed version of Hadoop on the cluster. |
| server-params | A comma-separated list of hadoop config parameters specified key-value pairs. These will be used to generate a hadoop-site.xml that will be used by the NameNode and DataNodes. |
| final-server-params | Same as above, except they will be marked final. |
gridservice-mapred options
| Option Name | Description |
|---|---|
| external |
If false, this indicates that a MapReduce cluster must be bought up by the HOD system on the nodes which it allocates via the allocate command. If true, if will try and connect to an externally configured MapReduce system. |
| host | Hostname of the externally configured JobTracker, if any. |
| tracker_port | Port to which the JobTracker RPC server is bound. |
| info_port | Port to which the JobTracker web UI server is bound. |
| pkgs | Installation directory, under which bin/hadoop executable is located. This can be used to use a pre-installed version of Hadoop on the cluster. |
| server-params | A comma-separated list of hadoop config parameters specified key-value pairs. These will be used to generate a hadoop-site.xml that will be used by the JobTracker and TaskTrackers. |
| final-server-params | Same as above, except they will be marked final. |