org.apache.spark.internal.io

Class FileCommitProtocol

Object

org.apache.spark.internal.io.FileCommitProtocol

Direct Known Subclasses:

HadoopMapReduceCommitProtocol

public abstract class FileCommitProtocol
extends Object

An interface to define how a single Spark job commits its outputs. Two notes:

1. Implementations must be serializable, as the committer instance instantiated on the driver will be used for tasks on executors. 2. Implementations should have a constructor with either 2 or 3 arguments: (jobId: String, path: String) or (jobId: String, path: String, isAppend: Boolean). 3. A committer should not be reused across multiple Spark jobs.

The proper call sequence is:

1. Driver calls setupJob. 2. As part of each task's execution, executor calls setupTask and then commitTask (or abortTask if task failed). 3. When all necessary tasks completed successfully, the driver calls commitJob. If the job failed to execute (e.g. too many failed tasks), the job should call abortJob.

Nested Class Summary

Nested Classes

Modifier and Type	Class and Description
static class	FileCommitProtocol.EmptyTaskCommitMessage$
static class	FileCommitProtocol.TaskCommitMessage

Constructor Summary

Constructors

Constructor and Description
FileCommitProtocol()

Method Summary

Methods

Modifier and Type	Method and Description
abstract void	abortJob(org.apache.hadoop.mapreduce.JobContext jobContext) Aborts a job after the writes fail.
abstract void	abortTask(org.apache.hadoop.mapreduce.TaskAttemptContext taskContext) Aborts a task after the writes have failed.
abstract void	commitJob(org.apache.hadoop.mapreduce.JobContext jobContext, scala.collection.Seq<FileCommitProtocol.TaskCommitMessage> taskCommits) Commits a job after the writes succeed.
abstract FileCommitProtocol.TaskCommitMessage	commitTask(org.apache.hadoop.mapreduce.TaskAttemptContext taskContext) Commits a task after the writes succeed.
static FileCommitProtocol	instantiate(String className, String jobId, String outputPath, boolean isAppend) Instantiates a FileCommitProtocol using the given className.
abstract String	newTaskTempFile(org.apache.hadoop.mapreduce.TaskAttemptContext taskContext, scala.Option<String> dir, String ext) Notifies the commit protocol to add a new file, and gets back the full path that should be used.
abstract String	newTaskTempFileAbsPath(org.apache.hadoop.mapreduce.TaskAttemptContext taskContext, String absoluteDir, String ext) Similar to newTaskTempFile(), but allows files to committed to an absolute output location.
abstract void	setupJob(org.apache.hadoop.mapreduce.JobContext jobContext) Setups up a job.
abstract void	setupTask(org.apache.hadoop.mapreduce.TaskAttemptContext taskContext) Sets up a task within a job.

Methods inherited from class Object

equals, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Constructor Detail

FileCommitProtocol

public FileCommitProtocol()

Method Detail

instantiate

public static FileCommitProtocol instantiate(String className,
                             String jobId,
                             String outputPath,
                             boolean isAppend)

Instantiates a FileCommitProtocol using the given className.

Parameters:

className - (undocumented)

jobId - (undocumented)

outputPath - (undocumented)

isAppend - (undocumented)

Returns:

(undocumented)

setupJob

public abstract void setupJob(org.apache.hadoop.mapreduce.JobContext jobContext)

Setups up a job. Must be called on the driver before any other methods can be invoked.

Parameters:

jobContext - (undocumented)

commitJob

public abstract void commitJob(org.apache.hadoop.mapreduce.JobContext jobContext,
             scala.collection.Seq<FileCommitProtocol.TaskCommitMessage> taskCommits)

Commits a job after the writes succeed. Must be called on the driver.

Parameters:

jobContext - (undocumented)

taskCommits - (undocumented)

abortJob

public abstract void abortJob(org.apache.hadoop.mapreduce.JobContext jobContext)

Aborts a job after the writes fail. Must be called on the driver.

Calling this function is a best-effort attempt, because it is possible that the driver just crashes (or killed) before it can call abort.

Parameters:

jobContext - (undocumented)

setupTask

public abstract void setupTask(org.apache.hadoop.mapreduce.TaskAttemptContext taskContext)

Sets up a task within a job. Must be called before any other task related methods can be invoked.

Parameters:

taskContext - (undocumented)

newTaskTempFile

public abstract String newTaskTempFile(org.apache.hadoop.mapreduce.TaskAttemptContext taskContext,
                     scala.Option<String> dir,
                     String ext)

Notifies the commit protocol to add a new file, and gets back the full path that should be used. Must be called on the executors when running tasks.

Note that the returned temp file may have an arbitrary path. The commit protocol only promises that the file will be at the location specified by the arguments after job commit.

A full file path consists of the following parts: 1. the base path 2. some sub-directory within the base path, used to specify partitioning 3. file prefix, usually some unique job id with the task id 4. bucket id 5. source specific file extension, e.g. ".snappy.parquet"

The "dir" parameter specifies 2, and "ext" parameter specifies both 4 and 5, and the rest are left to the commit protocol implementation to decide.

Important: it is the caller's responsibility to add uniquely identifying content to "ext" if a task is going to write out multiple files to the same dir. The file commit protocol only guarantees that files written by different tasks will not conflict.

Parameters:

taskContext - (undocumented)

dir - (undocumented)

ext - (undocumented)

Returns:

(undocumented)

newTaskTempFileAbsPath

public abstract String newTaskTempFileAbsPath(org.apache.hadoop.mapreduce.TaskAttemptContext taskContext,
                            String absoluteDir,
                            String ext)

Similar to newTaskTempFile(), but allows files to committed to an absolute output location. Depending on the implementation, there may be weaker guarantees around adding files this way.

Important: it is the caller's responsibility to add uniquely identifying content to "ext" if a task is going to write out multiple files to the same dir. The file commit protocol only guarantees that files written by different tasks will not conflict.

Parameters:

taskContext - (undocumented)

absoluteDir - (undocumented)

ext - (undocumented)

Returns:

(undocumented)

commitTask

public abstract FileCommitProtocol.TaskCommitMessage commitTask(org.apache.hadoop.mapreduce.TaskAttemptContext taskContext)

Commits a task after the writes succeed. Must be called on the executors when running tasks.

Parameters:

taskContext - (undocumented)

Returns:

(undocumented)

abortTask

public abstract void abortTask(org.apache.hadoop.mapreduce.TaskAttemptContext taskContext)

Aborts a task after the writes have failed. Must be called on the executors when running tasks.

Calling this function is a best-effort attempt, because it is possible that the executor just crashes (or killed) before it can call abort.

Parameters:

taskContext - (undocumented)