Swift Integration

Hadoop and Swift integration are the essential continuation of the Hadoop/OpenStack marriage. The key component to making this marriage work is the Hadoop Swift filesystem implementation. Although this implementation has been merged into the upstream Hadoop project, Sahara maintains a version with the most current features enabled.

Now the latest version of this jar (which uses Keystone API v3) is used in the plugins’ images automatically during build of these images. But for Ambari plugin we need to explicitly put this jar into /opt directory of the base image before cluster launching.

Hadoop patching

You may build the jar file yourself by choosing the latest patch from the Sahara Extra repository and using Maven to build with the pom.xml file provided. Or you may get the latest jar pre-built at https://tarballs.openstack.org/sahara-extra/dist/hadoop-openstack/master/

You will need to put this file into the hadoop libraries (e.g. /usr/lib/share/hadoop/lib, it depends on the plugin which you use) on each ResourceManager and NodeManager node (for Hadoop 2.x) in the cluster.

Hadoop configurations

In general, when Sahara runs a job on a cluster it will handle configuring the Hadoop installation. In cases where a user might require more in-depth configuration all the data is set in the core-site.xml file on the cluster instances using this template:

<property>
    <name>${name} + ${config}</name>
    <value>${value}</value>
    <description>${not mandatory description}</description>
</property>

There are two types of configs here:

  1. General. The ${name} in this case equals to fs.swift. Here is the list of ${config}:

    • .impl - Swift FileSystem implementation. The ${value} is org.apache.hadoop.fs.swift.snative.SwiftNativeFileSystem

    • .connect.timeout - timeout for all connections by default: 15000

    • .socket.timeout - how long the connection waits for responses from servers. by default: 60000

    • .connect.retry.count - connection retry count for all connections. by default: 3

    • .connect.throttle.delay - delay in millis between bulk (delete, rename, copy operations). by default: 0

    • .blocksize - blocksize for filesystem. By default: 32Mb

    • .partsize - the partition size for uploads. By default: 4608*1024Kb

    • .requestsize - request size for reads in KB. By default: 64Kb

  2. Provider-specific. The patch for Hadoop supports different cloud providers. The ${name} in this case equals to fs.swift.service.${provider}.

    Here is the list of ${config}:

    • .auth.url - authorization URL

    • .auth.endpoint.prefix - prefix for the service url, e.g. /AUTH_

    • .tenant - project name

    • .username

    • .password

    • .domain.name - Domains can be used to specify users who are not in the project specified.

    • .domain.id - You can also specify domain using id.

    • .trust.id - Trusts are optionally used to scope the authentication tokens of the supplied user.

    • .http.port

    • .https.port

    • .region - Swift region is used when cloud has more than one Swift installation. If region param is not set first region from Keystone endpoint list will be chosen. If region param not found exception will be thrown.

    • .location-aware - turn On location awareness. Is false by default

    • .apikey

    • .public

Example

For this example it is assumed that you have setup a Hadoop instance with a valid configuration and the Swift filesystem component. Furthermore there is assumed to be a Swift container named integration holding an object named temp, as well as a Keystone user named admin with a password of swordfish.

The following example illustrates how to copy an object to a new location in the same container. We will use Hadoop’s distcp command (http://hadoop.apache.org/docs/stable/hadoop-distcp/DistCp.html) to accomplish the copy. Note that the service provider for our Swift access is sahara, and that we will not need to specify the project of our Swift container as it will be provided in the Hadoop configuration.

Swift paths are expressed in Hadoop according to the following template: swift://${container}.${provider}/${object}. For our example source this will appear as swift://integration.sahara/temp.

Let’s run the job:

$ hadoop distcp -D fs.swift.service.sahara.username=admin \
 -D fs.swift.service.sahara.password=swordfish \
 swift://integration.sahara/temp swift://integration.sahara/temp1

After that just confirm that temp1 has been created in our integration container.

Limitations

Note: Please note that container names should be a valid URI.