Configuration Settings

Slack Docker Pulls GitHub edit source

An Alluxio cluster can be configured by setting the values of Alluxio configuration properties. The two major components to configure are Alluxio servers, consisting of masters and workers, and Alluxio clients, which are a part of applications.

Configure Applications

Customizing how an application job interacts with Alluxio service is specific to each application. The following are recommendations for some common applications.

Note that it is only valid to set client-side configurations for applications, such as properties prefixed with alluxio.user. Setting server-side properties on the application has no effect, suchj as properties prefixed with either alluxio.master or alluxio.worker.

Alluxio Shell Commands

Alluxio shell users can put JVM system properties -Dproperty=value after the fs command and before the subcommand to specify Alluxio user properties from the command line. For example, the following Alluxio shell command sets the write type to CACHE_THROUGH when copying files to Alluxio:

bin/alluxio fs -Dalluxio.user.file.writetype.default=CACHE_THROUGH copyFromLocal README.md /README.md

Note that, as a part of Alluxio deployment, Alluxio shell will also take the configuration in ${ALLUXIO_HOME}/conf/alluxio-site.properties when it is run from Alluxio installation at ${ALLUXIO_HOME}.

Spark

To customize Alluxio client-side properties in Spark, Spark users can use pass Alluxio properties as JVM system properties. See examples for the entire Spark Service or for individual Spark Jobs.

Hadoop MapReduce

See examples to configure Alluxio properties for the entire MapReduce service or for individual MapReduce jobs.

Hive

Hive can be configured to use customized Alluxio client-side properties for the entire service. See examples.

Presto

Presto can be configured to use customized Alluxio client-side properties for the entire service. See examples.

Configure Alluxio Cluster

Alluxio admins can create and customize the property file alluxio-site.properties to configure an Alluxio master or worker. If this file does not exist, it can be created from the template file under ${ALLUXIO_HOME}/conf:

cp conf/alluxio-site.properties.template conf/alluxio-site.properties

Make sure that this file is distributed to ${ALLUXIO_HOME}/conf on every Alluxio master and worker before starting the cluster. Any updates to the server configuration requires a restart of the process.

Use Environment Variables

Alluxio supports a few frequently used configuration settings via the environment variables, including:

Environment VariableDescription
ALLUXIO_CONF_DIR path to Alluxio configuration directory.
ALLUXIO_LOGS_DIR path to Alluxio logs directory.
ALLUXIO_MASTER_HOSTNAME hostname of Alluxio master, defaults to localhost.
ALLUXIO_UNDERFS_ADDRESS under storage system address, defaults to ${ALLUXIO_HOME}/underFSStorage which is a local file system.
ALLUXIO_RAM_FOLDER the directory where a worker stores in-memory data, defaults to /mnt/ramdisk.
ALLUXIO_JAVA_OPTS Java VM options for both Master, Worker and Alluxio Shell configuration. Note that, by default ALLUXIO_JAVA_OPTS is included in both ALLUXIO_MASTER_JAVA_OPTS, ALLUXIO_WORKER_JAVA_OPTS and ALLUXIO_USER_JAVA_OPTS.
ALLUXIO_MASTER_JAVA_OPTS additional Java VM options for Master configuration.
ALLUXIO_WORKER_JAVA_OPTS additional Java VM options for Worker configuration.
ALLUXIO_USER_JAVA_OPTS additional Java VM options for Alluxio shell configuration.
ALLUXIO_CLASSPATH additional classpath entries for Alluxio processes. This is empty by default.
ALLUXIO_LOGSERVER_HOSTNAME host name of the log server. This is empty by default.
ALLUXIO_LOGSERVER_PORT port number of the log server. This is 45600 by default.
ALLUXIO_LOGSERVER_LOGS_DIR path to the local directory where Alluxio log server stores logs received from Alluxio servers.

For example, to setup the following:

  • an Alluxio master at localhost
  • the root mount point as an HDFS cluster with a namenode also running at localhost
  • enable Java remote debugging at port 7001 run the following commands before startingthe master process:
export ALLUXIO_MASTER_HOSTNAME="localhost"
export ALLUXIO_UNDERFS_ADDRESS="hdfs://localhost:9000"
export ALLUXIO_MASTER_JAVA_OPTS="$ALLUXIO_JAVA_OPTS -agentlib:jdwp=transport=dt_socket,server=y,suspend=n,address=7001"

Users can either set these variables through the shell or in conf/alluxio-env.sh. If this file does not exist yet, create one by copying the template:

cp conf/alluxio-env.sh.template conf/alluxio-env.sh

Specify Cluster-Wide Defaults

Since version 1.8, each Alluxio client or worker can initialize its configuration with the cluster-wide configuration values retrieved from masters.

When different client applications (Alluxio Shell CLI, Spark jobs, MapReduce jobs) or Alluxio workers connect to an Alluxio master, they will initialize their own Alluxio configuration properties with the default values supplied by the masters based on the master-side ${ALLUXIO_HOME}/conf/alluxio-site.properties files. As a result, cluster admins can set client-side settings (e.g., alluxio.user.*), or network transport settings (e.g., alluxio.security.authentication.type), or worker settings (e.g., alluxio.worker.*) in ${ALLUXIO_HOME}/conf/alluxio-site.properties on all the masters, which will be distributed and become cluster-wide default values when clients and workers connect.

For example, the property alluxio.user.file.writetype.default defaults to MUST_CACHE, which only writes to Alluxio space. In an Alluxio cluster where data persistence is preferred and all jobs need to write to both the UFS and Alluxio, the administrator can add alluxio.user.file.writetype.default=CACHE_THROUGH in each master’s alluxio-site.properties file. After restarting the cluster, all jobs will automatically set alluxio.user.file.writetype.default to CACHE_THROUGH.

Clients can ignore or overwrite the cluster-wide default values by either specifying the user property alluxio.user.conf.cluster.default.enabled=false to decline loading the cluster-wide default values or following the approaches described in Configure Alluxio for Applications to overwrite the same properties.

Note that, before version 1.8, ${ALLUXIO_HOME}/conf/alluxio-site.properties file is only loaded by Alluxio server processes and will be ignored by applications interacting with Alluxio service through Alluxio client, unless ${ALLUXIO_HOME}/conf is on applications’ classpath.

Configuration Sources

An Alluxio property can be possibly configured in multiple sources. In this case, its final value is determined by the following priority list, from highest priority to lowest:

  1. JVM system properties (i.e., -Dproperty=key)
  2. Environment variables
  3. Property files: When an Alluxio cluster starts, each server process including master and worker searches for alluxio-site.properties within the following directories in the given order, stopping when a match is found: ${HOME}/.alluxio/, /etc/alluxio/, and ${ALLUXIO_HOME}/conf
  4. Cluster default values: An Alluxio client may initialize its configuration based on the cluster-wide default configuration served by the masters.

If no user-specified configuration is found for a property, Alluxio runtime will fallback to its default property value.

To check the value of a specific configuration property and the source of its value, users can run the following command:

bin/alluxio getConf alluxio.worker.port
29998
bin/alluxio getConf --source alluxio.worker.port
DEFAULT

To list all of the configuration properties with sources:

bin/alluxio getConf --source
alluxio.conf.dir=/Users/bob/alluxio/conf (SYSTEM_PROPERTY)
alluxio.debug=false (DEFAULT)
...

Users can also specify the --master option to list all of the cluster-wide configuration properties served by the masters. Note that with the --master option, getConf will query the master, requiring the master nodes be running. Otherwise, without --master option, this command only checks the local configuration.

bin/alluxio getConf --master --source
alluxio.conf.dir=/Users/bob/alluxio/conf (SYSTEM_PROPERTY)
alluxio.debug=false (DEFAULT)
...

Server Configuration Checker

The server-side configuration checker helps discover configuration errors and warnings. Suspected configuration errors are reported through the web UI, doctor CLI, and master logs.

The web UI shows the result of the server configuration check.

webUi

Users can also run the fsadmin doctor command to get the same results.

bin/alluxio fsadmin doctor configuration

Configuration warnings can also be found in the master logs.

masterLogs