public class ZombieJob extends Object implements JobStory
ZombieJob
is a layer above LoggedJob
raw JSON objects.
Each ZombieJob
object represents a job in job history. For everything
that exists in job history, contents are returned unchanged faithfully. To
get input splits of a non-exist task, a non-exist task attempt, or an
ill-formed task attempt, proper objects are made up from statistical
sketches.构造器和说明 |
---|
ZombieJob(LoggedJob job,
ClusterStory cluster)
|
ZombieJob(LoggedJob job,
ClusterStory cluster,
long seed)
|
限定符和类型 | 方法和说明 |
---|---|
InputSplit[] |
getInputSplits()
Get the input splits for the job.
|
JobConf |
getJobConf()
Get the
JobConf for the job. |
JobID |
getJobID()
Get the job ID
|
LoggedJob |
getLoggedJob()
Get the underlining
LoggedJob object read directly from the trace. |
TaskAttemptInfo |
getMapTaskAttemptInfoAdjusted(int taskNumber,
int taskAttemptNumber,
int locality)
Get a
TaskAttemptInfo with a TaskAttemptID associated with
taskType, taskNumber, and taskAttemptNumber. |
String |
getName()
Get the job name.
|
int |
getNumberMaps()
Get the number of maps in the
JobStory . |
int |
getNumberReduces()
Get the number of reduce in the
JobStory . |
int |
getNumLoggedMaps()
Getting the number of map tasks that are actually logged in the trace.
|
int |
getNumLoggedReduces()
Getting the number of reduce tasks that are actually logged in the trace.
|
Pre21JobHistoryConstants.Values |
getOutcome()
Get the outcome of the job execution.
|
String |
getQueueName()
Get the queue where the job is submitted.
|
long |
getSubmissionTime()
Get the job submission time.
|
TaskAttemptInfo |
getTaskAttemptInfo(TaskType taskType,
int taskNumber,
int taskAttemptNumber)
Get a
TaskAttemptInfo with a TaskAttemptID associated with
taskType, taskNumber, and taskAttemptNumber. |
TaskInfo |
getTaskInfo(TaskType taskType,
int taskNumber)
Get
TaskInfo for a given task. |
String |
getUser()
Get the user who ran the job.
|
public ZombieJob(LoggedJob job, ClusterStory cluster, long seed)
ZombieJob
with the same semantics as the
LoggedJob
passed in this parameterjob
- The dead job this ZombieJob instance is based on.cluster
- The cluster topology where the dead job ran on. This argument can
be null if we do not have knowledge of the cluster topology.seed
- Seed for the random number generator for filling in information
not available from the ZombieJob.public ZombieJob(LoggedJob job, ClusterStory cluster)
ZombieJob
with the same semantics as the
LoggedJob
passed in this parameterjob
- The dead job this ZombieJob instance is based on.cluster
- The cluster topology where the dead job ran on. This argument can
be null if we do not have knowledge of the cluster topology.public JobConf getJobConf()
JobStory
JobConf
for the job.getJobConf
在接口中 JobStory
JobConf
for the jobpublic InputSplit[] getInputSplits()
JobStory
getInputSplits
在接口中 JobStory
public int getNumberMaps()
JobStory
JobStory
.getNumberMaps
在接口中 JobStory
Job
public int getNumberReduces()
JobStory
JobStory
.getNumberReduces
在接口中 JobStory
Job
public Pre21JobHistoryConstants.Values getOutcome()
JobStory
getOutcome
在接口中 JobStory
public long getSubmissionTime()
JobStory
getSubmissionTime
在接口中 JobStory
public String getQueueName()
JobStory
getQueueName
在接口中 JobStory
public int getNumLoggedMaps()
public int getNumLoggedReduces()
public LoggedJob getLoggedJob()
LoggedJob
object read directly from the trace.
This is mainly for debugging.LoggedJob
objectpublic TaskAttemptInfo getTaskAttemptInfo(TaskType taskType, int taskNumber, int taskAttemptNumber)
TaskAttemptInfo
with a TaskAttemptID
associated with
taskType, taskNumber, and taskAttemptNumber. This function does not care
about locality, and follows the following decision logic: 1. Make up a
TaskAttemptInfo
if the task attempt is missing in trace, 2. Make up
a TaskAttemptInfo
if the task attempt has a KILLED final status in
trace, 3. Otherwise (final state is SUCCEEDED or FAILED), construct the
TaskAttemptInfo
from the trace.getTaskAttemptInfo
在接口中 JobStory
taskType
- TaskType
of the task-attempttaskNumber
- Partition number of the task-attempttaskAttemptNumber
- Attempt number of the taskTaskAttemptInfo
for the given task-attemptpublic TaskInfo getTaskInfo(TaskType taskType, int taskNumber)
JobStory
TaskInfo
for a given task.getTaskInfo
在接口中 JobStory
taskType
- TaskType
of the tasktaskNumber
- Partition number of the taskTaskInfo
for the given taskpublic TaskAttemptInfo getMapTaskAttemptInfoAdjusted(int taskNumber, int taskAttemptNumber, int locality)
TaskAttemptInfo
with a TaskAttemptID
associated with
taskType, taskNumber, and taskAttemptNumber. This function considers
locality, and follows the following decision logic: 1. Make up a
TaskAttemptInfo
if the task attempt is missing in trace, 2. Make up
a TaskAttemptInfo
if the task attempt has a KILLED final status in
trace, 3. If final state is FAILED, construct a TaskAttemptInfo
from the trace, without considering locality. 4. If final state is
SUCCEEDED, construct a TaskAttemptInfo
from the trace, with runtime
scaled according to locality in simulation and locality in trace.getMapTaskAttemptInfoAdjusted
在接口中 JobStory
taskNumber
- Partition number of the task-attempttaskAttemptNumber
- Attempt number of the tasklocality
- Data locality of the task as scheduled in simulationTaskAttemptInfo
for the given task-attemptCopyright © 2009 The Apache Software Foundation