org.apache.hadoop.fs.s3a.commit.files

Class SuccessData

java.lang.Object

org.apache.hadoop.fs.s3a.commit.files.PersistentCommitData<SuccessData>

org.apache.hadoop.fs.s3a.commit.files.SuccessData

All Implemented Interfaces:

Serializable, org.apache.hadoop.fs.statistics.IOStatisticsSource

@InterfaceAudience.Private
 @InterfaceStability.Unstable
public class SuccessData
extends PersistentCommitData<SuccessData>

Summary data saved into a _SUCCESS marker file. This provides an easy way to determine which committer was used to commit work.

1. File length == 0: classic FileOutputCommitter.
2. Loadable as SuccessData: An S3A committer with name in in committer field.
3. Not loadable? Something else.

This should be considered public, and MUST stay compatible at the JSON format level with that of org.apache.hadoop.mapreduce.lib.output.committer.manifest.files.ManifestSuccessData

The JSON format SHOULD be considered public and evolving with compatibility across versions.

All the Java serialization data is different and may change across versions with no stability guarantees other than "manifest summaries MAY be serialized between processes with the exact same version of this binary on their classpaths." That is sufficient for testing in Spark.

To aid with Java serialization, the maps and lists are exclusively those which serialize well. IOStatisticsSnapshot has a lot of complexity in marshalling there; this class doesn't worry about concurrent access so is simpler. Note: to deal with scale issues, the S3A committers do not include any more than the number of objects listed in CommitConstants.SUCCESS_MARKER_FILE_LIMIT. This is intended to suffice for basic integration tests. Larger tests should examine the generated files themselves.

See Also:

Serialized Form

Field Summary

Fields

Modifier and Type	Field and Description
static String	NAME Name to include in persisted data, so as to differentiate from any other manifests: "org.apache.hadoop.fs.s3a.commit.files.SuccessData/1".
static int	VERSION Supported version value: 1.

Constructor Summary

Constructors

Constructor and Description
SuccessData()

Method Summary

Modifier and Type	Method and Description
void	addDiagnostic(String key, String value) Add a diagnostics entry.
String	dumpDiagnostics(String prefix, String middle, String suffix) Dump the diagnostics (if any) to a string.
String	dumpMetrics(String prefix, String middle, String suffix) Dump the metrics (if any) to a string.
String	getCommitter()
String	getDate()
String	getDescription()
Map<String,String>	getDiagnostics()
List<String>	getFilenames()
String	getHostname()
org.apache.hadoop.fs.statistics.IOStatisticsSnapshot	getIOStatistics()
String	getJobId()
String	getJobIdSource()
Map<String,Long>	getMetrics()
String	getName()
String	getStage()
String	getState()
boolean	getSuccess() Get the success flag.
long	getTimestamp()
protected static String	joinMap(Map<String,?> map, String prefix, String middle, String suffix) Join any map of string to value into a string, sorting the keys first.
static SuccessData	load(org.apache.hadoop.fs.FileSystem fs, org.apache.hadoop.fs.Path path) Load an instance from a file, then validate it.
void	putDiagnostic(String key, String value) Add a diagnostics entry.
void	recordJobFailure(Throwable thrown) Note a failure by setting success flag to false, then add the exception to the diagnostics.
org.apache.hadoop.fs.statistics.IOStatistics	save(org.apache.hadoop.fs.FileSystem fs, org.apache.hadoop.fs.Path path, org.apache.hadoop.util.JsonSerialization<SuccessData> serializer) Save to a hadoop filesystem.
static org.apache.hadoop.util.JsonSerialization<SuccessData>	serializer() Get a JSON serializer for this class.
void	setCommitter(String committer)
void	setDate(String date)
void	setDescription(String description)
void	setDiagnostics(Map<String,String> diagnostics)
void	setFilenames(List<String> filenames)
void	setHostname(String hostname)
void	setIOStatistics(org.apache.hadoop.fs.statistics.IOStatisticsSnapshot ioStatistics)
void	setJobId(String jobId)
void	setJobIdSource(String jobIdSource)
void	setMetrics(Map<String,Long> metrics)
void	setName(String name)
void	setState(String state)
void	setSuccess(boolean success) Set the success flag.
void	setTimestamp(long timestamp)
byte[]	toBytes(org.apache.hadoop.util.JsonSerialization<SuccessData> serializer) Serialize to JSON and then to a byte array, after performing a preflight validation of the data to be saved.
String	toString()
void	validate() Validate the data: those fields which must be non empty, must be set.

Methods inherited from class org.apache.hadoop.fs.s3a.commit.files.PersistentCommitData

load, saveFile, saveToStream

Methods inherited from class java.lang.Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, wait, wait, wait

Field Detail

VERSION

public static final int VERSION

Supported version value: 1. If this is changed the value of serialVersionUID will change, to avoid deserialization problems.

See Also:

Constant Field Values

NAME

public static final String NAME

Name to include in persisted data, so as to differentiate from any other manifests: "org.apache.hadoop.fs.s3a.commit.files.SuccessData/1".

See Also:

Constant Field Values

Constructor Detail

SuccessData

public SuccessData()

Method Detail

validate

public void validate()
              throws ValidationFailure

Description copied from class: PersistentCommitData

Validate the data: those fields which must be non empty, must be set.

Specified by:

validate in class PersistentCommitData<SuccessData>

Throws:

ValidationFailure - if the data is invalid

toBytes

public byte[] toBytes(org.apache.hadoop.util.JsonSerialization<SuccessData> serializer)
               throws IOException

Description copied from class: PersistentCommitData

Serialize to JSON and then to a byte array, after performing a preflight validation of the data to be saved.

Specified by:

toBytes in class PersistentCommitData<SuccessData>

Parameters:

serializer - serializer to use

Returns:

the data in a persistable form.

Throws:

IOException - serialization problem or validation failure.

save

public org.apache.hadoop.fs.statistics.IOStatistics save(org.apache.hadoop.fs.FileSystem fs,
                                                         org.apache.hadoop.fs.Path path,
                                                         org.apache.hadoop.util.JsonSerialization<SuccessData> serializer)
                                                  throws IOException

Description copied from class: PersistentCommitData

Save to a hadoop filesystem. The destination file is overwritten, and on s3a stores the performance flag is set to turn off all existence checks and parent dir cleanup. The assumption here is: the job knows what it is doing.

Specified by:

save in class PersistentCommitData<SuccessData>

Parameters:

fs - filesystem

path - path

serializer - serializer to use

Returns:

IOStats from the output stream.

Throws:

IOException - IO exception

toString

public String toString()

Overrides:

toString in class Object

dumpMetrics

public String dumpMetrics(String prefix,
                          String middle,
                          String suffix)

Dump the metrics (if any) to a string. The metrics are sorted for ease of viewing.

Parameters:

prefix - prefix before every entry

middle - string between key and value

suffix - suffix to each entry

Returns:

the dumped string

dumpDiagnostics

public String dumpDiagnostics(String prefix,
                              String middle,
                              String suffix)

Dump the diagnostics (if any) to a string.

Parameters:

prefix - prefix before every entry

middle - string between key and value

suffix - suffix to each entry

Returns:

the dumped string

joinMap

protected static String joinMap(Map<String,?> map,
                                String prefix,
                                String middle,
                                String suffix)

Join any map of string to value into a string, sorting the keys first.

Parameters:

map - map to join

prefix - prefix before every entry

middle - string between key and value

suffix - suffix to each entry

Returns:

a string for reporting.

load

public static SuccessData load(org.apache.hadoop.fs.FileSystem fs,
                               org.apache.hadoop.fs.Path path)
                        throws IOException

Load an instance from a file, then validate it.

Parameters:

fs - filesystem

path - path

Returns:

the loaded instance

Throws:

IOException - IO failure

ValidationFailure - if the data is invalid

serializer

public static org.apache.hadoop.util.JsonSerialization<SuccessData> serializer()

Get a JSON serializer for this class.

Returns:

a serializer.

getName

public String getName()

setName

public void setName(String name)

getTimestamp

public long getTimestamp()

Returns:

timestamp of creation.

setTimestamp

public void setTimestamp(long timestamp)

getDate

public String getDate()

Returns:

timestamp as date; no expectation of parseability.

setDate

public void setDate(String date)

getHostname

public String getHostname()

Returns:

host which created the file (implicitly: committed the work).

setHostname

public void setHostname(String hostname)

getCommitter

public String getCommitter()

Returns:

committer name.

setCommitter

public void setCommitter(String committer)

getDescription

public String getDescription()

Returns:

any description text.

setDescription

public void setDescription(String description)

getMetrics

public Map<String,Long> getMetrics()

Returns:

any metrics.

setMetrics

public void setMetrics(Map<String,Long> metrics)

getFilenames

public List<String> getFilenames()

Returns:

a list of filenames in the commit.

setFilenames

public void setFilenames(List<String> filenames)

getDiagnostics

public Map<String,String> getDiagnostics()

setDiagnostics

public void setDiagnostics(Map<String,String> diagnostics)

addDiagnostic

public void addDiagnostic(String key,
                          String value)

Add a diagnostics entry.

Parameters:

key - name

value - value

getJobId

public String getJobId()

Returns:

Job ID, if known.

setJobId

public void setJobId(String jobId)

getJobIdSource

public String getJobIdSource()

setJobIdSource

public void setJobIdSource(String jobIdSource)

getIOStatistics

public org.apache.hadoop.fs.statistics.IOStatisticsSnapshot getIOStatistics()

setIOStatistics

public void setIOStatistics(org.apache.hadoop.fs.statistics.IOStatisticsSnapshot ioStatistics)

setSuccess

public void setSuccess(boolean success)

Set the success flag.

Parameters:

success - did the job succeed?

getSuccess

public boolean getSuccess()

Get the success flag.

Returns:

did the job succeed?

getState

public String getState()

setState

public void setState(String state)

getStage

public String getStage()

putDiagnostic

public void putDiagnostic(String key,
                          String value)

Add a diagnostics entry.

Parameters:

key - name

value - value

recordJobFailure

public void recordJobFailure(Throwable thrown)

Note a failure by setting success flag to false, then add the exception to the diagnostics.

Parameters:

thrown - throwable