Elastic Data Processing (EDP)

Overview

Sahara’s Elastic Data Processing facility or EDP allows the execution of jobs on clusters created from sahara. EDP supports:

• Hive, Pig, MapReduce, MapReduce.Streaming, Java, and Shell job types on Hadoop clusters

• Spark jobs on Spark standalone clusters, MapR (v5.0.0 - v5.2.0) clusters, Vanilla clusters (v2.7.1) and CDH clusters (v5.3.0 or higher).

• storage of job binaries in the OpenStack Object Storage service (swift), the OpenStack Shared file systems service (manila), sahara’s own database, or any S3-like object store

• access to input and output data sources in

  • HDFS for all job types

  • swift for all types excluding Hive

  • manila (NFS shares only) for all types excluding Pig

  • Any S3-like object store

• configuration of jobs at submission time

• execution of jobs on existing clusters or transient clusters

Interfaces

The EDP features can be used from the sahara web UI which is described in the Sahara (Data Processing) UI User Guide.

The EDP features also can be used directly by a client through the REST api

EDP Concepts

Sahara EDP uses a collection of simple objects to define and execute jobs. These objects are stored in the sahara database when they are created, allowing them to be reused. This modular approach with database persistence allows code and data to be reused across multiple jobs.

The essential components of a job are:

• executable code to run

• input and output data paths, as needed for the job

• any additional configuration values needed for the job run

These components are supplied through the objects described below.

Job Binaries

A Job Binary object stores a URL to a single script or Jar file and any credentials needed to retrieve the file. The file itself may be stored in the sahara internal database (only API v1.1), in swift, or in manila.

Files in the sahara database are stored as raw bytes in a Job Binary Internal object. This object’s sole purpose is to store a file for later retrieval. No extra credentials need to be supplied for files stored internally.

Sahara requires credentials (username and password) to access files stored in swift unless swift proxy users are configured as described in Sahara Advanced Configuration Guide. The swift service must be running in the same OpenStack installation referenced by sahara.

Sahara requires the following credentials/configs to access files stored in an S3-like object store: accesskey, secretkey, endpoint. These credentials are specified through the extra in the body of the request when creating a job binary referencing S3. The value of endpoint should include a protocol: http or https.

To reference a binary file stored in manila, create the job binary with the URL manila://{share_id}/{path}. This assumes that you have already stored that file in the appropriate path on the share. The share will be automatically mounted to any cluster nodes which require access to the file, if it is not mounted already.

There is a configurable limit on the size of a single job binary that may be retrieved by sahara. This limit is 5MB and may be set with the job_binary_max_KB setting in the sahara.conf configuration file.

Jobs

A Job object specifies the type of the job and lists all of the individual Job Binary objects that are required for execution. An individual Job Binary may be referenced by multiple Jobs. A Job object specifies a main binary and/or supporting libraries depending on its type:

Job type	Main binary	Libraries
Hive	required	optional
Pig	required	optional
MapReduce	not used	required
MapReduce.Streaming	not used	optional
Java	not used	required
Shell	required	optional
Spark	required	optional
Storm	required	not used
Storm Pyelus	required	not used

Data Sources

A Data Source object stores a URL which designates the location of input or output data and any credentials needed to access the location.

Sahara supports data sources in swift. The swift service must be running in the same OpenStack installation referenced by sahara.

Sahara also supports data sources in HDFS. Any HDFS instance running on a sahara cluster in the same OpenStack installation is accessible without manual configuration. Other instances of HDFS may be used as well provided that the URL is resolvable from the node executing the job.

Sahara supports data sources in manila as well. To reference a path on an NFS share as a data source, create the data source with the URL manila://{share_id}/{path}. As in the case of job binaries, the specified share will be automatically mounted to your cluster’s nodes as needed to access the data source.

Finally, Sahara supports data sources referring to S3-like object stores. The URL should be of the form s3://{bucket}/{path}. Also, the following credentials/configs are understood: accesskey, secretkey, endpoint, bucket_in_path, and ssl. These credentials are specified through the credentials attribute of the body of the request when creating a data source referencing S3. The value of endpoint should NOT include a protocol (http or https), unlike when referencing an S3 job binary. It can also be noted that Sahara clusters can interact with S3-like stores even when not using EDP, i.e. when manually operating the cluster instead. Consult the hadoop-aws documentation for more information. Also, be advised that hadoop-aws will only write a job’s output into a bucket which already exists: it does not create new buckets.

Some job types require the use of data source objects to specify input and output when a job is launched. For example, when running a Pig job the UI will prompt the user for input and output data source objects.

Other job types like Java or Spark do not require the user to specify data sources. For these job types, data paths are passed as arguments. For convenience, sahara allows data source objects to be referenced by name or id. The section Using Data Source References as Arguments gives further details.

Job Execution

Job objects must be launched or executed in order for them to run on the cluster. During job launch, a user specifies execution details including data sources, configuration values, and program arguments. The relevant details will vary by job type. The launch will create a Job Execution object in sahara which is used to monitor and manage the job.

To execute Hadoop jobs, sahara generates an Oozie workflow and submits it to the Oozie server running on the cluster. Familiarity with Oozie is not necessary for using sahara but it may be beneficial to the user. A link to the Oozie web console can be found in the sahara web UI in the cluster details.

For Spark jobs, sahara uses the spark-submit shell script and executes the Spark job from the master node in case of Spark cluster and from the Spark Job History server in other cases. Logs of spark jobs run by sahara can be found on this node under the /tmp/spark-edp directory.

General Workflow

The general workflow for defining and executing a job in sahara is essentially the same whether using the web UI or the REST API.

1. Launch a cluster from sahara if there is not one already available

2. Create all of the Job Binaries needed to run the job, stored in the sahara database, in swift, or in manila

   • When using the REST API and internal storage of job binaries, the Job Binary Internal objects must be created first

   • Once the Job Binary Internal objects are created, Job Binary objects may be created which refer to them by URL

3. Create a Job object which references the Job Binaries created in step 2

4. Create an input Data Source which points to the data you wish to process

5. Create an output Data Source which points to the location for output data

6. Create a Job Execution object specifying the cluster and Job object plus relevant data sources, configuration values, and program arguments

   • When using the web UI this is done with the Launch On Existing Cluster or Launch on New Cluster buttons on the Jobs tab

   • When using the REST API this is done via the /jobs/<job_id>/execute method

The workflow is simpler when using existing objects. For example, to construct a new job which uses existing binaries and input data a user may only need to perform steps 3, 5, and 6 above. Of course, to repeat the same job multiple times a user would need only step 6.

Specifying Configuration Values, Parameters, and Arguments

Jobs can be configured at launch. The job type determines the kinds of values that may be set:

Job type	Configuration Values	Parameters	Arguments
Hive	Yes	Yes	No
Pig	Yes	Yes	Yes
MapReduce	Yes	No	No
MapReduce.Streaming	Yes	No	No
Java	Yes	No	Yes
Shell	Yes	Yes	Yes
Spark	Yes	No	Yes
Storm	Yes	No	Yes
Storm Pyelus	Yes	No	Yes

• Configuration values are key/value pairs.

  • The EDP configuration values have names beginning with edp. and are consumed by sahara

  • Other configuration values may be read at runtime by Hadoop jobs

  • Currently additional configuration values are not available to Spark jobs at runtime

• Parameters are key/value pairs. They supply values for the Hive and Pig parameter substitution mechanisms. In Shell jobs, they are passed as environment variables.

• Arguments are strings passed as command line arguments to a shell or main program

These values can be set on the Configure tab during job launch through the web UI or through the job_configs parameter when using the /jobs/<job_id>/execute REST method.

In some cases sahara generates configuration values or parameters automatically. Values set explicitly by the user during launch will override those generated by sahara.

Using Data Source References as Arguments

Sometimes it’s necessary or desirable to pass a data path as an argument to a job. In these cases, a user may simply type out the path as an argument when launching a job. If the path requires credentials, the user can manually add the credentials as configuration values. However, if a data source object has been created that contains the desired path and credentials there is no need to specify this information manually.

As a convenience, sahara allows data source objects to be referenced by name or id in arguments, configuration values, or parameters. When the job is executed, sahara will replace the reference with the path stored in the data source object and will add any necessary credentials to the job configuration. Referencing an existing data source object is much faster than adding this information by hand. This is particularly useful for job types like Java or Spark that do not use data source objects directly.

There are two job configuration parameters that enable data source references. They may be used with any job type and are set on the Configuration tab when the job is launched:

• edp.substitute_data_source_for_name (default False) If set to True, causes sahara to look for data source object name references in configuration values, arguments, and parameters when a job is launched. Name references have the form datasource://name_of_the_object.

  For example, assume a user has a WordCount application that takes an input path as an argument. If there is a data source object named my_input, a user may simply set the edp.substitute_data_source_for_name configuration parameter to True and add datasource://my_input as an argument when launching the job.

• edp.substitute_data_source_for_uuid (default False) If set to True, causes sahara to look for data source object ids in configuration values, arguments, and parameters when a job is launched. A data source object id is a uuid, so they are unique. The id of a data source object is available through the UI or the sahara command line client. A user may simply use the id as a value.

Creating an Interface for Your Job

In order to better document your job for cluster operators (or for yourself in the future), sahara allows the addition of an interface (or method signature) to your job template. A sample interface for the Teragen Hadoop example might be:

Name	Mapping Type	Location	Value Type	Required	Default
Example Class	args	0	string	false	teragen
Rows	args	1	number	true	unset
Output Path	args	2	data_source	false	hdfs://ip:port/path
Mapper Count	configs	mapred. map.tasks	number	false	unset

A “Description” field may also be added to each interface argument.

To create such an interface via the REST API, provide an “interface” argument, the value of which consists of a list of JSON objects, as below:

[
    {
        "name": "Example Class",
        "description": "Indicates which example job class should be used.",
        "mapping_type": "args",
        "location": "0",
        "value_type": "string",
        "required": false,
        "default": "teragen"
    },
]

Creating this interface would allow you to specify a configuration for any execution of the job template by passing an “interface” map similar to:

{
    "Rows": "1000000",
    "Mapper Count": "3",
    "Output Path": "hdfs://mycluster:8020/user/myuser/teragen-output"
}

The specified arguments would be automatically placed into the args, configs, and params for the job, according to the mapping type and location fields of each interface argument. The final job_configs map would be:

{
    "job_configs": {
        "configs":
            {
                "mapred.map.tasks": "3"
            },
        "args":
            [
                "teragen",
                "1000000",
                "hdfs://mycluster:8020/user/myuser/teragen-output"
            ]
    }
}

Rules for specifying an interface are as follows:

• Mapping Type must be one of configs, params, or args. Only types supported for your job type are allowed (see above.)

• Location must be a string for configs and params, and an integer for args. The set of args locations must be an unbroken series of integers starting from 0.

• Value Type must be one of string, number, or data_source. Data sources may be passed as UUIDs or as valid paths (see above.) All values should be sent as JSON strings. (Note that booleans and null values are serialized differently in different languages. Please specify them as a string representation of the appropriate constants for your data processing engine.)

• args that are not required must be given a default value.

The additional one-time complexity of specifying an interface on your template allows a simpler repeated execution path, and also allows us to generate a customized form for your job in the Horizon UI. This may be particularly useful in cases in which an operator who is not a data processing job developer will be running and administering the jobs.

Generation of Swift Properties for Data Sources

If swift proxy users are not configured (see Sahara Advanced Configuration Guide) and a job is run with data source objects containing swift paths, sahara will automatically generate swift username and password configuration values based on the credentials in the data sources. If the input and output data sources are both in swift, it is expected that they specify the same credentials.

The swift credentials may be set explicitly with the following configuration values:

Name
fs.swift.service.sahara.username
fs.swift.service.sahara.password

Setting the swift credentials explicitly is required when passing literal swift paths as arguments instead of using data source references. When possible, use data source references as described in Using Data Source References as Arguments.

Additional Details for Hive jobs

Sahara will automatically generate values for the INPUT and OUTPUT parameters required by Hive based on the specified data sources.

Additional Details for Pig jobs

Sahara will automatically generate values for the INPUT and OUTPUT parameters required by Pig based on the specified data sources.

For Pig jobs, arguments should be thought of as command line arguments separated by spaces and passed to the pig shell.

Parameters are a shorthand and are actually translated to the arguments -param name=value

Additional Details for MapReduce jobs

Important!

If the job type is MapReduce, the mapper and reducer classes must be specified as configuration values.

Note that the UI will not prompt the user for these required values; they must be added manually with the Configure tab.

Make sure to add these values with the correct names:

Name	Example Value
mapred.mapper.new-api	true
mapred.reducer.new-api	true
mapreduce.job.map.class	org.apache.oozie.example.SampleMapper
mapreduce.job.reduce.class	org.apache.oozie.example.SampleReducer

Additional Details for MapReduce.Streaming jobs

Important!

If the job type is MapReduce.Streaming, the streaming mapper and reducer classes must be specified.

In this case, the UI will prompt the user to enter mapper and reducer values on the form and will take care of adding them to the job configuration with the appropriate names. If using the python client, however, be certain to add these values to the job configuration manually with the correct names:

Name	Example Value
edp.streaming.mapper	/bin/cat
edp.streaming.reducer	/usr/bin/wc

Additional Details for Java jobs

Data Source objects are not used directly with Java job types. Instead, any input or output paths must be specified as arguments at job launch either explicitly or by reference as described in Using Data Source References as Arguments. Using data source references is the recommended way to pass paths to Java jobs.

If configuration values are specified, they must be added to the job’s Hadoop configuration at runtime. There are two methods of doing this. The simplest way is to use the edp.java.adapt_for_oozie option described below. The other method is to use the code from this example to explicitly load the values.

The following special configuration values are read by sahara and affect how Java jobs are run:

• edp.java.main_class (required) Specifies the full name of the class containing main(String[] args)

  A Java job will execute the main method of the specified main class. Any arguments set during job launch will be passed to the program through the args array.

• oozie.libpath (optional) Specifies configuration values for the Oozie share libs, these libs can be shared by different workflows

• edp.java.java_opts (optional) Specifies configuration values for the JVM

• edp.java.adapt_for_oozie (optional) Specifies that sahara should perform special handling of configuration values and exit conditions. The default is False.

  If this configuration value is set to True, sahara will modify the job’s Hadoop configuration before invoking the specified main method. Any configuration values specified during job launch (excluding those beginning with edp.) will be automatically set in the job’s Hadoop configuration and will be available through standard methods.

  Secondly, setting this option to True ensures that Oozie will handle program exit conditions correctly.

At this time, the following special configuration value only applies when running jobs on a cluster generated by the Cloudera plugin with the Enable Hbase Common Lib cluster config set to True (the default value):

• edp.hbase_common_lib (optional) Specifies that a common Hbase lib generated by sahara in HDFS be added to the oozie.libpath. This for use when an Hbase application is driven from a Java job. Default is False.

The edp-wordcount example bundled with sahara shows how to use configuration values, arguments, and swift data paths in a Java job type. Note that the example does not use the edp.java.adapt_for_oozie option but includes the code to load the configuration values explicitly.

Additional Details for Shell jobs

A shell job will execute the script specified as main, and will place any files specified as libs in the same working directory (on both the filesystem and in HDFS). Command line arguments may be passed to the script through the args array, and any params values will be passed as environment variables.

Data Source objects are not used directly with Shell job types but data source references may be used as described in Using Data Source References as Arguments.

The edp-shell example bundled with sahara contains a script which will output the executing user to a file specified by the first command line argument.

Additional Details for Spark jobs

Data Source objects are not used directly with Spark job types. Instead, any input or output paths must be specified as arguments at job launch either explicitly or by reference as described in Using Data Source References as Arguments. Using data source references is the recommended way to pass paths to Spark jobs.

Spark jobs use some special configuration values:

• edp.java.main_class (required) Specifies the full name of the class containing the Java or Scala main method:

  • main(String[] args) for Java

  • main(args: Array[String] for Scala

  A Spark job will execute the main method of the specified main class. Any arguments set during job launch will be passed to the program through the args array.

• edp.spark.adapt_for_swift (optional) If set to True, instructs sahara to modify the job’s Hadoop configuration so that swift paths may be accessed. Without this configuration value, swift paths will not be accessible to Spark jobs. The default is False. Despite the name, the same principle applies to jobs which reference paths in S3-like stores.

• edp.spark.driver.classpath (optional) If set to empty string sahara will use default classpath for the cluster during job execution. Otherwise this will override default value for the cluster for particular job execution.

The edp-spark example bundled with sahara contains a Spark program for estimating Pi.

Special Sahara URLs

Sahara uses custom URLs to refer to objects stored in swift, in manila, in the sahara internal database, or in S3-like storage. These URLs are usually not meant to be used outside of sahara.

Sahara swift URLs passed to running jobs as input or output sources include a “.sahara” suffix on the container, for example:

swift://container.sahara/object

You may notice these swift URLs in job logs, however, you do not need to add the suffix to the containers yourself. sahara will add the suffix if necessary, so when using the UI or the python client you may write the above URL simply as:

swift://container/object

Sahara internal database URLs have the form:

internal-db://sahara-generated-uuid

This indicates a file object in the sahara database which has the given uuid as a key.

Manila NFS filesystem reference URLS take the form:

manila://share-uuid/path

This format should be used when referring to a job binary or a data source stored in a manila NFS share.

For both job binaries and data sources, S3 urls take the form:

s3://bucket/path/to/object

Despite the above URL format, the current implementation of EDP will still use the Hadoop s3a driver to access data sources. Botocore is used to access job binaries.

EDP Requirements

The OpenStack installation and the cluster launched from sahara must meet the following minimum requirements in order for EDP to function:

OpenStack Services

When a Hadoop job is executed, binaries are first uploaded to a cluster node and then moved from the node local filesystem to HDFS. Therefore, there must be an instance of HDFS available to the nodes in the sahara cluster.

If the swift service is not running in the OpenStack installation:

• Job binaries may only be stored in the sahara internal database

• Data sources require a long-running HDFS

If the swift service is running in the OpenStack installation:

• Job binaries may be stored in swift or the sahara internal database

• Data sources may be in swift or a long-running HDFS

Cluster Processes

Requirements for EDP support depend on the EDP job type and plugin used for the cluster. For example a Vanilla sahara cluster must run at least one instance of these processes to support EDP:

• For Hadoop version 1:

  • jobtracker

  • namenode

  • oozie

  • tasktracker

  • datanode

• For Hadoop version 2:

  • namenode

  • datanode

  • resourcemanager

  • nodemanager

  • historyserver

  • oozie

  • spark history server

EDP Technical Considerations

There are several things in EDP which require attention in order to work properly. They are listed on this page.

Transient Clusters

EDP allows running jobs on transient clusters. In this case the cluster is created specifically for the job and is shut down automatically once the job is finished.

Two config parameters control the behaviour of periodic clusters:

• periodic_enable - if set to ‘false’, sahara will do nothing to a transient

  cluster once the job it was created for is completed. If it is set to ‘true’, then the behaviour depends on the value of the next parameter.

• use_identity_api_v3 - set it to ‘false’ if your OpenStack installation

  does not provide keystone API v3. In that case sahara will not terminate unneeded clusters. Instead it will set their state to ‘AwaitingTermination’ meaning that they could be manually deleted by a user. If the parameter is set to ‘true’, sahara will itself terminate the cluster. The limitation is caused by lack of ‘trusts’ feature in Keystone API older than v3.

If both parameters are set to ‘true’, sahara works with transient clusters in the following manner:

1. When a user requests for a job to be executed on a transient cluster, sahara creates such a cluster.

2. Sahara drops the user’s credentials once the cluster is created but prior to that it creates a trust allowing it to operate with the cluster instances in the future without user credentials.

3. Once a cluster is not needed, sahara terminates its instances using the stored trust. sahara drops the trust after that.