::Go back to Oozie Documentation Index::


Oozie Web Services API, V1 (Workflow, Coordinator, And Bundle)

The Oozie Web Services API is a HTTP REST JSON API.

All responses are in UTF-8 .

Assuming Oozie is runing at OOZIE_URL , the following web services end points are supported:

  • /versions
  • /v1/admin
  • /v1/job
  • /v1/jobs
  • /v2/job
  • /v2/jobs
  • /v2/admin
  • /v2/sla

Documentation on the API is below; in some cases, looking at the corresponding command in the Command Line Documentation page will provide additional details and examples. Most of the functionality offered by the Oozie CLI is using the WS API. If you export OOZIE_DEBUG then the Oozie CLI will output the WS API details used by any commands you execute. This is useful for debugging purposes to or see how the Oozie CLI works with the WS API.

Versions End-Point

Identical to the corresponding Oozie v0 WS API

This endpoint is for clients to perform protocol negotiation.

It support only HTTP GET request and not sub-resources.

It returns the supported Oozie protocol versions by the server.

Current returned values are 0, 1, 2 .

Request:

GET /oozie/versions

Response:

HTTP/1.1 200 OK
Content-Type: application/json;charset=UTF-8
.
[0,1]

Admin End-Point

This endpoint is for obtaining Oozie system status and configuration information.

It supports the following sub-resources: status, os-env, sys-props, configuration, instrumentation, systems, available-timezones .

System Status

Identical to the corresponding Oozie v0 WS API

A HTTP GET request returns the system status.

Request:

GET /oozie/v1/admin/status

Response:

HTTP/1.1 200 OK
Content-Type: application/json;charset=UTF-8
.
{"systemMode":NORMAL}

With a HTTP PUT request it is possible to change the system status between NORMAL , NOWEBSERVICE , and SAFEMODE .

Request:

PUT /oozie/v1/admin/status?systemmode=SAFEMODE

Response:

HTTP/1.1 200 OK

OS Environment

Identical to the corresponding Oozie v0 WS API

A HTTP GET request returns the Oozie system OS environment.

Request:

GET /oozie/v1/admin/os-env

Response:

HTTP/1.1 200 OK
Content-Type: application/json;charset=UTF-8
.
{
  TERM: "xterm",
  JAVA_HOME: "/usr/java/latest",
  XCURSOR_SIZE: "",
  SSH_CLIENT: "::ffff:127.0.0.1 49082 22",
  XCURSOR_THEME: "default",
  INPUTRC: "/etc/inputrc",
  HISTSIZE: "1000",
  PATH: "/usr/java/latest/bin"
  KDE_FULL_SESSION: "true",
  LANG: "en_US.UTF-8",
  ...
}

Java System Properties

Identical to the corresponding Oozie v0 WS API

A HTTP GET request returns the Oozie Java system properties.

Request:

GET /oozie/v1/admin/java-sys-properties

Response:

HTTP/1.1 200 OK
Content-Type: application/json;charset=UTF-8
.
{
  java.vm.version: "11.0-b15",
  sun.jnu.encoding: "UTF-8",
  java.vendor.url: "http://java.sun.com/",
  java.vm.info: "mixed mode",
  ...
}

Oozie Configuration

Identical to the corresponding Oozie v0 WS API

A HTTP GET request returns the Oozie system configuration.

Request:

GET /oozie/v1/admin/configuration

Response:

HTTP/1.1 200 OK
Content-Type: application/json;charset=UTF-8
.
{
  oozie.service.SchedulerService.threads: "5",
  oozie.service.ActionService.executor.classes: "
            org.apache.oozie.dag.action.decision.DecisionActionExecutor,
            org.apache.oozie.dag.action.hadoop.HadoopActionExecutor,
            org.apache.oozie.dag.action.hadoop.FsActionExecutor
        ",
  oozie.service.CallableQueueService.threads.min: "10",
  oozie.service.DBLiteWorkflowStoreService.oozie.autoinstall: "true",
  ...
}

Oozie Instrumentation

Identical to the corresponding Oozie v0 WS API

A HTTP GET request returns the Oozie instrumentation information.

Request:

GET /oozie/v1/admin/instrumentation

Response:

HTTP/1.1 200 OK
Content-Type: application/json;charset=UTF-8
.
{
  timers: [
    {
      group: "db",
      data: [
        {
          ownMinTime: 2,
          ownTimeStdDev: 0,
          totalTimeStdDev: 0,
          ownTimeAvg: 3,
          ticks: 117,
          name: "update-workflow",
          ownMaxTime: 32,
          totalMinTime: 2,
          totalMaxTime: 32,
          totalTimeAvg: 3
        },
        ...
      ]
    },
    ...
  ],
  samplers: [
    {
      group: "callablequeue",
      data: [
        {
          name: "threads.active",
          value: 1.8333333333333333
        },
        {
          name: "delayed.queue.size",
          value: 0
        },
        {
          name: "queue.size",
          value: 0
        }
      ]
    },
    ...
  ],
  variables: [
    {
      group: "jvm",
      data: [
        {
          name: "max.memory",
          value: 506920960
        },
        {
          name: "total.memory",
          value: 56492032
        },
        {
          name: "free.memory",
          value: 45776800
        }
      ]
    },
    ...
  ]
}

Version

Identical to the corresponding Oozie v0 WS API

A HTTP GET request returns the Oozie build version.

Request:

GET /oozie/v1/admin/build-version

Response:

HTTP/1.1 200 OK
Content-Type: application/json;charset=UTF-8
.
{buildVersion: "3.0.0-SNAPSHOT" }

Available Time Zones

A HTTP GET request returns the available time zones.

Request:

GET /oozie/v1/admin/available-timezones

Response:

HTTP/1.1 200 OK
Content-Type: application/json;charset=UTF-8
.
{
  "available-timezones":[
    {
      "timezoneDisplayName":"SST (Pacific\/Midway)",
      "timezoneId":"Pacific\/Midway"
    },
    {
      "timezoneDisplayName":"NUT (Pacific\/Niue)",
      "timezoneId":"Pacific\/Niue"
    },
    {
      "timezoneDisplayName":"SST (Pacific\/Pago_Pago)",
      "timezoneId":"Pacific\/Pago_Pago"
    },
    {
      "timezoneDisplayName":"SST (Pacific\/Samoa)",
      "timezoneId":"Pacific\/Samoa"
    },
    {
      "timezoneDisplayName":"SST (US\/Samoa)",
      "timezoneId":"US\/Samoa"
    },
    {
      "timezoneDisplayName":"HAST (America\/Adak)",
      "timezoneId":"America\/Adak"
    },
    {
      "timezoneDisplayName":"HAST (America\/Atka)",
      "timezoneId":"America\/Atka"
    },
    {
      "timezoneDisplayName":"HST (HST)",
      "timezoneId":"HST"
    },
    ...
  ]
}

Queue Dump

A HTTP GET request returns the queue dump of the Oozie system. This is an administrator debugging feature.

Request:

GET /oozie/v1/admin/queue-dump

Available Oozie Servers

A HTTP GET request returns the list of available Oozie Servers. This is useful when Oozie is configured for High Availability ; if not, it will simply return the one Oozie Server.

Request:

GET /oozie/v2/admin/available-oozie-servers

Response:

HTTP/1.1 200 OK
Content-Type: application/json;charset=UTF-8
.
{
    "hostA": "http://hostA:11000/oozie",
    "hostB": "http://hostB:11000/oozie",
    "hostC": "http://hostC:11000/oozie",
}

List available sharelib

A HTTP GET request to get list of available sharelib. If the name of the sharelib is passed as an argument (regex supported) then all corresponding files are also listed.

Request:

GET /oozie/v2/admin/list_sharelib

Response:

HTTP/1.1 200 OK
Content-Type: application/json;charset=UTF-8
{
    "sharelib":
        [
            "oozie",
            "hive",
            "distcp",
            "hcatalog",
            "sqoop",
            "mapreduce-streaming",
            "pig"
        ]
}

Request:

GET /oozie/v2/admin/list_sharelib?lib=pig*

Response:

HTTP/1.1 200 OK
Content-Type: application/json;charset=UTF-8{
    "sharelib":
            [
                {
                    "pig":
                        {
                            "sharelibFiles":
                                [
                                    hdfs://localhost:9000/user/purushah/share/lib/lib_20131114095729/pig/pig.jar
                                    hdfs://localhost:9000/user/purushah/share/lib/lib_20131114095729/pig/piggybank.jar
                                ]
                        }
                }
            ]
}

Update system sharelib

This webservice call makes the oozie server(s) to pick up the latest version of sharelib present under oozie.service.WorkflowAppService.system.libpath directory based on the sharelib directory timestamp or reloads the sharelib metafile if one is configured. The main purpose is to update the sharelib on the oozie server without restarting.

Request:

GET /oozie/v2/admin/update_sharelib

Response:

HTTP/1.1 200 OK
Content-Type: application/json;charset=UTF-8
[
    {
       "sharelibUpdate":{
          "host":"server1",
          "status":"Server not found"
       }
    },
    {
       "sharelibUpdate":{
          "host":"server2",
          "status":"Successful",
          "sharelibDirOld":"hdfs://localhost:51951/user/purushah/share/lib/lib_20140107181218",
          "sharelibDirNew":"hdfs://localhost:51951/user/purushah/share/lib/lib_20140107181218"
       }
    },
    {
       "sharelibUpdate":{
          "host":"server3",
          "status":"Successful",
          "sharelibDirOld":"hdfs://localhost:51951/user/purushah/share/lib/lib_20140107181218",
          "sharelibDirNew":"hdfs://localhost:51951/user/purushah/share/lib/lib_20140107181218"
       }
    }
]

Job and Jobs End-Points

Modified in Oozie v1 WS API

These endpoints are for submitting, managing and retrieving information of workflow, coordinator, and bundle jobs.

Job Submission
Standard Job Submission

An HTTP POST request with an XML configuration as payload creates a job.

The type of job is determined by the presence of one of the following 3 properties:

  • oozie.wf.application.path : path to a workflow aplication directory, creates a workflow job
  • oozie.coord.application.path : path to a coordinator application file, creates a coordinator job
  • oozie.bundle.application.path : path to a bundle application file, creates a bundle job

Or, if none of those are present, the jobtype parameter determines the type of job to run. It can either be mapreduce or pig.

Request:

POST /oozie/v1/jobs
Content-Type: application/xml;charset=UTF-8
.
<?xml version="1.0" encoding="UTF-8"?>
<configuration>
    <property>
        <name>user.name</name>
        <value>bansalm</value>
    </property>
    <property>
        <name>oozie.wf.application.path</name>
        <value>hdfs://foo:8020/user/bansalm/myapp/</value>
    </property>
    ...
</configuration>

Response:

HTTP/1.1 201 CREATED
Content-Type: application/json;charset=UTF-8
.
{
  id: "job-3"
}

A created job will be in PREP status. If the query string parameter 'action=start' is provided in the POST URL, the job will be started immediately and its status will be RUNNING .

Coordinator jobs with start time in the future they will not create any action until the start time happens.

A coordinator job will remain in PREP status until it's triggered, in which case it will change to RUNNING status. The 'action=start' parameter is not valid for coordinator jobs.

Proxy MapReduce Job Submission

You can submit a Workflow that contains a single MapReduce action without writing a workflow.xml. Any required Jars or other files must already exist in HDFS.

The following properties are required; any additional parameters needed by the MapReduce job can also be specified here:

  • fs.default.name : The NameNode
  • mapred.job.tracker : The JobTracker
  • mapred.mapper.class : The map-task classname
  • mapred.reducer.class : The reducer-task classname
  • mapred.input.dir : The map-task input directory
  • mapred.output.dir : The reduce-task output directory
  • user.name : The username of the user submitting the job
  • oozie.libpath : A directory in HDFS that contains necessary Jars for your job
  • oozie.proxysubmission : Must be set to true

Request:

POST /oozie/v1/jobs?jobtype=mapreduce
Content-Type: application/xml;charset=UTF-8
.
<?xml version="1.0" encoding="UTF-8"?>
<configuration>
    <property>
        <name>fs.default.name</name>
        <value>hdfs://localhost:8020</value>
    </property>
    <property>
        <name>mapred.job.tracker</name>
        <value>localhost:8021</value>
    </property>
    <property>
        <name>mapred.mapper.class</name>
        <value>org.apache.oozie.example.SampleMapper</value>
    </property>
    <property>
        <name>mapred.reducer.class</name>
        <value>org.apache.oozie.example.SampleReducer</value>
    </property>
    <property>
        <name>mapred.input.dir</name>
        <value>hdfs://localhost:8020/user/rkanter/examples/input-data/text</value>
    </property>
    <property>
        <name>mapred.output.dir</name>
        <value>hdfs://localhost:8020/user/rkanter/examples/output-data/map-reduce</value>
    </property>
    <property>
        <name>user.name</name>
        <value>rkanter</value>
    </property>
    <property>
        <name>oozie.libpath</name>
        <value>hdfs://localhost:8020/user/rkanter/examples/apps/map-reduce/lib</value>
    </property>
    <property>
        <name>oozie.proxysubmission</name>
        <value>true</value>
    </property>
</configuration>

Response:

HTTP/1.1 201 CREATED
Content-Type: application/json;charset=UTF-8
.
{
  id: "job-3"
}

Proxy Pig Job Submission

You can submit a Workflow that contains a single Pig action without writing a workflow.xml. Any requred Jars or other files must already exist in HDFS.

The following properties are required:

  • fs.default.name : The NameNode
  • mapred.job.tracker : The JobTracker
  • user.name : The username of the user submitting the job
  • oozie.pig.script : Contains the pig script you want to run (the actual script, not a file path)
  • oozie.libpath : A directory in HDFS that contains necessary Jars for your job
  • oozie.proxysubmission : Must be set to true

The following properties are optional:

  • oozie.pig.script.params.size : The number of parameters you'll be passing to Pig
  • oozie.pig.script.params.n : A parameter (variable definition for the script) in 'key=value' format, the 'n' should be an integer starting with 0 to indicate the parameter number
  • oozie.pig.options.size : The number of options you'll be passing to Pig
  • oozie.pig.options.n : An argument to pass to Pig, the 'n' should be an integer starting with 0 to indicate the option number

The oozie.pig.options.n parameters are sent directly to Pig without any modification unless they start with -D , in which case they are put into the element of the action.

In addition to passing parameters to Pig with oozie.pig.script.params.n , you can also create a properties file on HDFS and reference it with the -param_file option in oozie.pig.script.options.n ; both are shown in the following example.

$ hadoop fs -cat /user/rkanter/pig_params.properties
INPUT=/user/rkanter/examples/input-data/text

Request:

POST /oozie/v1/jobs?jobtype=pig
Content-Type: application/xml;charset=UTF-8
.
<?xml version="1.0" encoding="UTF-8"?>
<configuration>
    <property>
        <name>fs.default.name</name>
        <value>hdfs://localhost:8020</value>
    </property>
    <property>
        <name>mapred.job.tracker</name>
        <value>localhost:8021</value>
    </property>
    <property>
        <name>user.name</name>
        <value>rkanter</value>
    </property>
    <property>
        <name>oozie.pig.script</name>
        <value>
            A  load '$INPUT' using PigStorage(':');
            B = foreach A generate $0 as id;
            store B into '$OUTPUT' USING PigStorage();
        </value>
    </property>
    <property>
        <name>oozie.pig.script.params.size</name>
        <value>1</value>
    </property>
    <property>
        <name>oozie.pig.script.params.0</name>
        <value>OUTPUT
/user/rkanter/examples/output-data/pig</value>
    </property>
    <property>
        <name>oozie.pig.options.size</name>
        <value>2</value>
    </property>
    <property>
        <name>oozie.pig.options.0</name>
        <value>-param_file</value>
    </property>
    <property>
        <name>oozie.pig.options.1</name>
        <value>hdfs://localhost:8020/user/rkanter/pig_params.properties</value>
    </property>
    <property>
        <name>oozie.libpath</name>
        <value>hdfs://localhost:8020/user/rkanter/share/lib/pig</value>
    </property>
    <property>
        <name>oozie.proxysubmission</name>
        <value>true</value>
    </property>
</configuration>

Response:

HTTP/1.1 201 CREATED
Content-Type: application/json;charset=UTF-8
.
{
  id: "job-3"
}

Proxy Hive Job Submission

You can submit a Workflow that contains a single Hive action without writing a workflow.xml. Any requred Jars or other files must already exist in HDFS.

The following properties are required:

  • fs.default.name : The NameNode
  • mapred.job.tracker : The JobTracker
  • user.name : The username of the user submitting the job
  • oozie.hive.script : Contains the hive script you want to run (the actual script, not a file path)
  • oozie.libpath : A directory in HDFS that contains necessary Jars for your job
  • oozie.proxysubmission : Must be set to true

The following properties are optional:

  • oozie.hive.script.params.size : The number of parameters you'll be passing to Hive
  • oozie.hive.script.params.n : A parameter (variable definition for the script) in 'key=value' format, the 'n' should be an integer starting with 0 to indicate the parameter number
  • oozie.hive.options.size : The number of options you'll be passing to Pig
  • oozie.hive.options.n : An argument to pass to Hive, the 'n' should be an integer starting with 0 to indicate the option number

The oozie.hive.options.n parameters are sent directly to Hive without any modification unless they start with -D , in which case they are put into the element of the action.

Request:

POST /oozie/v1/jobs?jobtype=hive
Content-Type: application/xml;charset=UTF-8
.
<?xml version="1.0" encoding="UTF-8"?>
<configuration>
    <property>
        <name>fs.default.name</name>
        <value>hdfs://localhost:8020</value>
    </property>
    <property>
        <name>mapred.job.tracker</name>
        <value>localhost:8021</value>
    </property>
    <property>
        <name>user.name</name>
        <value>rkanter</value>
    </property>
    <property>
        <name>oozie.hive.script</name>
        <value>
            CREATE EXTERNAL TABLE test (a INT) STORED AS TEXTFILE LOCATION '${INPUT}';
            INSERT OVERWRITE DIRECTORY '${OUTPUT}' SELECT * FROM test;
        </value>
    </property>
    <property>
        <name>oozie.hive.script.params.size</name>
        <value>2</value>
    </property>
    <property>
        <name>oozie.hive.script.params.0</name>
        <value>OUTPUT=/user/rkanter/examples/output-data/hive</value>
    </property>
    <property>
        <name>oozie.hive.script.params.1</name>
        <value>INPUT=/user/rkanter/examples/input-data/table</value>
    </property>
    <property>
        <name>oozie.libpath</name>
        <value>hdfs://localhost:8020/user/rkanter/share/lib/hive</value>
    </property>
    <property>
        <name>oozie.proxysubmission</name>
        <value>true</value>
    </property>
</configuration>

Response:

HTTP/1.1 201 CREATED
Content-Type: application/json;charset=UTF-8
.
{
  id: "job-3"
}

Proxy Sqoop Job Submission

You can submit a Workflow that contains a single Sqoop command without writing a workflow.xml. Any requred Jars or other files must already exist in HDFS.

The following properties are required:

  • fs.default.name : The NameNode
  • mapred.job.tracker : The JobTracker
  • user.name : The username of the user submitting the job
  • oozie.sqoop.command : The sqoop command you want to run where each argument occupies one line or separated by "\n"
  • oozie.libpath : A directory in HDFS that contains necessary Jars for your job
  • oozie.proxysubmission : Must be set to true

The following properties are optional:

  • oozie.sqoop.options.size : The number of options you'll be passing to Sqoop Hadoop job
  • oozie.sqoop.options.n : An argument to pass to Sqoop hadoop job conf, the 'n' should be an integer starting with 0 to indicate the option number

Request:

POST /oozie/v1/jobs?jobtype=sqoop
Content-Type: application/xml;charset=UTF-8
.
<?xml version="1.0" encoding="UTF-8"?>
<configuration>
    <property>
        <name>fs.default.name</name>
        <value>hdfs://localhost:8020</value>
    </property>
    <property>
        <name>mapred.job.tracker</name>
        <value>localhost:8021</value>
    </property>
    <property>
        <name>user.name</name>
        <value>bzhang</value>
    </property>
    <property>
        <name>oozie.sqoop.command</name>
        <value>
            import
            --connect
            jdbc:mysql://localhost:3306/oozie
            --username
            oozie
            --password
            oozie
            --table
            WF_JOBS
            --target-dir
            /user/${wf:user()}/${examplesRoot}/output-data/sqoop
        </value>
    </property>
        <name>oozie.libpath</name>
        <value>hdfs://localhost:8020/user/bzhang/share/lib/sqoop</value>
    </property>
    <property>
        <name>oozie.proxysubmission</name>
        <value>true</value>
    </property>
</configuration>

Response:

HTTP/1.1 201 CREATED
Content-Type: application/json;charset=UTF-8
.
{
  id: "job-3"
}

Managing a Job

A HTTP PUT request starts, suspends, resumes, kills, or dryruns a job.

Request:

PUT /oozie/v1/job/job-3?action=start

Response:

HTTP/1.1 200 OK

Valid values for the 'action' parameter are 'start', 'suspend', 'resume', 'kill', 'dryrun', 'rerun', and 'change'.

Rerunning and changing a job require additional parameters, and are described below:

Re-Runing a Workflow Job

A workflow job in SUCCEEDED , KILLED or FAILED status can be partially rerun specifying a list of workflow nodes to skip during the rerun. All the nodes in the skip list must have complete its execution.

The rerun job will have the same job ID.

A rerun request is done with a HTTP PUT request with a rerun action.

Request:

PUT /oozie/v1/job/job-3?action=rerun
Content-Type: application/xml;charset=UTF-8
.
<?xml version="1.0" encoding="UTF-8"?>
<configuration>
    <property>
        <name>user.name</name>
        <value>tucu</value>
    </property>
    <property>
        <name>oozie.wf.application.path</name>
        <value>hdfs://foo:8020/user/tucu/myapp/</value>
    </property>
    <property>
        <name>oozie.wf.rerun.skip.nodes</name>
        <value>firstAction,secondAction</value>
    </property>
    ...
</configuration>

Response:

HTTP/1.1 200 OK

Re-Runing a coordinator job

A coordinator job in RUNNING SUCCEEDED , KILLED or FAILED status can be partially rerun by specifying the coordinator actions to re-execute.

A rerun request is done with an HTTP PUT request with a coord-rerun action .

The type of the rerun can be date or action .

The scope of the rerun depends on the type: * date : a comma-separated list of date ranges. Each date range element is specified with dates separated by :: * action : a comma-separated list of action ranges. Each action range is specified with two action numbers separated by -

The refresh parameter can be true or false to specify if the user wants to refresh an action's input and output events.

The nocleanp paramter can be true or false to specify is the user wants to cleanup output events for the rerun actions.

Request:

PUT /oozie/v1/job/job-3?action=coord-rerun&type=action&scope=1-2&refresh=false&nocleanup=false
.

or

PUT /oozie/v1/job/job-3?action=coord-rerun&type=date2009-02-01T00:10Z::2009-03-01T00:10Z&scope=&refresh=false&nocleanup=false
.

Response:

HTTP/1.1 200 OK

Re-Runing a bundle job

A coordinator job in RUNNING SUCCEEDED , KILLED or FAILED status can be partially rerun by specifying the coordinators to re-execute.

A rerun request is done with an HTTP PUT request with a bundle-rerun action .

A comma separated list of coordinator job names (not IDs) can be specified in the coord-scope parameter.

The date-scope parameter is a comma-separated list of date ranges. Each date range element is specified with dates separated by :: . If empty or not included, Oozie will figure this out for you

The refresh parameter can be true or false to specify if the user wants to refresh the coordinator's input and output events.

The nocleanp paramter can be true or false to specify is the user wants to cleanup output events for the rerun coordinators.

Request:

PUT /oozie/v1/job/job-3?action=bundle-rerun&coord-scope=coord-1&refresh=false&nocleanup=false
.

Response:

HTTP/1.1 200 OK

Changing endtime/concurrency/pausetime of a Coordinator Job

A coordinator job not in KILLED status can have it's endtime, concurrency, or pausetime changed.

A change request is done with an HTTP PUT request with a change action .

The value parameter can contain any of the following: * endtime: the end time of the coordinator job. * concurrency: the concurrency of the coordinator job. * pausetime: the pause time of the coordinator job.

Multiple arguments can be passed to the value parameter by separating them with a ';' character.

If an already-succeeded job changes its end time, its status will become running.

Request:

PUT /oozie/v1/job/job-3?action=change&value=endtime=2011-12-01T05:00Z
.

or

PUT /oozie/v1/job/job-3?action=change&value=concurrency=100
.

or

PUT /oozie/v1/job/job-3?action=change&value=pausetime=2011-12-01T05:00Z
.

or

PUT /oozie/v1/job/job-3?action=change&value=endtime=2011-12-01T05:00Z;concurrency=100;pausetime=2011-12-01T05:00Z
.

Response:

HTTP/1.1 200 OK

Job Information

A HTTP GET request retrieves the job information.

Request:

GET /oozie/v1/job/job-3?show=info&timezone=GMT

Response for a workflow job:

HTTP/1.1 200 OK
Content-Type: application/json;charset=UTF-8
.
{ 
  id: "0-200905191240-oozie-W",
  appName: "indexer-workflow",
  appPath: "hdfs://user/bansalm/indexer.wf",
  externalId: "0-200905191230-oozie-pepe",
  user: "bansalm",
  status: "RUNNING",
  conf: "<configuration> ... </configuration>",
  createdTime: "Thu, 01 Jan 2009 00:00:00 GMT",
  startTime: "Fri, 02 Jan 2009 00:00:00 GMT",
  endTime: null,
  run: 0,
  actions: [
    {
      id: "0-200905191240-oozie-W@indexer",
      name: "indexer",
      type: "map-reduce",
      conf: "<configuration> ...</configuration>",
      startTime: "Thu, 01 Jan 2009 00:00:00 GMT",
      endTime: "Fri, 02 Jan 2009 00:00:00 GMT",
      status: "OK",
      externalId: "job-123-200903101010",
      externalStatus: "SUCCEEDED",
      trackerUri: "foo:8021",
      consoleUrl: "http://foo:50040/jobdetailshistory.jsp?jobId=...",
      transition: "reporter",
      data: null,
      errorCode: null,
      errorMessage: null,
      retries: 0
    },
    ...
  ]
}

Response for a coordinator job:

HTTP/1.1 200 OK
Content-Type: application/json;charset=UTF-8
.
{
  id: "0-200905191240-oozie-C",
  appName: "indexer-Coord",
  appPath: "hdfs://user/bansalm/myapp/logprocessor-coord.xml",
  externalId: "0-200905191230-oozie-pepe",
  user: "bansalm",
  status: "RUNNING",
  conf: "<configuration> ... </configuration>",
  createdTime: "Thu, 01 Jan 2009 00:00:00 GMT",
  startTime: "Fri, 02 Jan 2009 00:00:00 GMT",
  endTime: "Fri, 31 Dec 2009 00:00:00 GMT",
  frequency: "${days(1)}"
  actions: [
    {
      id: "0000010-130426111815091-oozie-bansalm-C@1",
      createdTime: "Fri, 26 Apr 2013 20:57:07 GMT",
      externalId: "",
      missingDependencies: "",
      runConf: null,
      createdConf: null,
      consoleUrl: null,
      nominalTime: "Fri, 01 Jan 2010 01:00:00 GMT",
  ...
}

Response for a bundle job:

HTTP/1.1 200 OK
Content-Type: application/json;charset=UTF-8
.
{
  jobType: "bundle",
  id: "0-200905191240-oozie-B",
  appName: "new-bundle",
  appPath: "hdfs://user/bansalm/myapp/logprocessor-bundle.xml",
  externalId: "0-200905191230-oozie-pepe",
  user: "bansalm",
  status: "RUNNING",
  conf: "<configuration> ... </configuration>",
  createdTime: "Thu, 01 Jan 2009 00:00:00 GMT",
  startTime: "Fri, 02 Jan 2009 00:00:00 GMT",
  endTime: "Fri, 31 Dec 2009 00:00:00 GMT"
  bundleCoordJobs: [
    {
      status: "RUNNING",
      concurrency: 1,
      conf: "<configuration> ... </configuration>",
      executionPolicy: "FIFO",
      toString: "Coordinator application id[0000010-130426111815091-oozie-bansalm-C] status[RUNNING]",
      coordJobName: "coord-1",
      endTime: "Fri, 01 Jan 2010 03:00:00 GMT",
      ...
    }
  ...
}

Getting all the Workflows corresponding to a Coordinator Action:

A coordinator action kicks off different workflows for its original run and all subsequent reruns. Getting a list of those workflow ids is a useful tool to keep track of your actions' runs and to go debug the workflow job logs if required. Along with ids, it also lists their statuses, and start and end times for quick reference.

Both v1 and v2 API are supported. v0 is not supported.

GET /oozie/v2/job/0000001-111219170928042-oozie-joe-C@1?show=allruns

Response

HTTP/1.1 200 OK
Content-Type: application/json;charset=UTF-8
.
{"workflows":[
    {
        "startTime":"Mon, 24 Mar 2014 23:40:53 GMT",
        "id":"0000001-140324163709596-oozie-chit-W",
        "status":"SUCCEEDED",
        "endTime":"Mon, 24 Mar 2014 23:40:54 GMT"
    },
    {
        "startTime":"Mon, 24 Mar 2014 23:44:01 GMT",
        "id":"0000000-140324164318985-oozie-chit-W",
        "status":"SUCCEEDED",
        "endTime":"Mon, 24 Mar 2014 23:44:01 GMT"
    },
    {
        "startTime":"Mon, 24 Mar 2014 23:44:24 GMT",
        "id":"0000001-140324164318985-oozie-chit-W",
        "status":"SUCCEEDED",
        "endTime":"Mon, 24 Mar 2014 23:44:24 GMT"
    }
]}

An alternate API is also available for the same output. With this API, one can pass the coordinator JOB Id followed by query params - type=action and scope=. One single action number can be passed at a time.

GET /oozie/v2/job/0000001-111219170928042-oozie-joe-C?show=allruns&type=action&scope=1

Retrieve a subset of actions

Query parameters, offset and length can be specified with a workflow job to retrieve specific actions. Default is offset=0, len=1000

GET /oozie/v1/job/0000002-130507145349661-oozie-joe-W?show=info&offset=5&len=10
Query parameters, offset , length , filter can be specified with a coordinator job to retrieve specific actions. Query parameter, order with value "desc" can be used to retrieve the latest coordinator actions materialized instead of actions from @1. Query parameters filter can be used to retrieve coodinator actions matching specific status. Default is offset=0, len=0 for v2/job (i.e., does not return any coordinator actions) and offset=0, len=1000 with v1/job and v0/job. So if you need actions to be returned with v2 API, specifying len parameter is necessary. Default order is "asc".
GET /oozie/v1/job/0000001-111219170928042-oozie-joe-C?show=info&offset=5&len=10&filter=status%3DKILLED&order=desc
Note that the filter is URL encoded, its decoded value is status=KILLED .
GET /oozie/v1/job/0000001-111219170928042-oozie-joe-C?show=info&filter=status%21%3DSUCCEEDED&order=desc
This retrives coordinator actions except for SUCCEEDED status, which is useful for debugging.

Job Application Definition

A HTTP GET request retrieves the workflow or a coordinator job definition file.

Request:

GET /oozie/v1/job/job-3?show=definition

Response for a workflow job:

HTTP/1.1 200 OK
Content-Type: application/xml;charset=UTF-8
.
<?xml version="1.0" encoding="UTF-8"?>
<workflow-app name='xyz-app' xmlns="uri:oozie:workflow:0.1">
    <start to='firstaction' />
    ...
    <end name='end' />
</workflow-app>

Response for a coordinator job:

HTTP/1.1 200 OK
Content-Type: application/xml;charset=UTF-8
.
<?xml version="1.0" encoding="UTF-8"?>
<coordinator-app name='abc-app' xmlns="uri:oozie:coordinator:0.1" frequency="${days(1)} 
                 start="2009-01-01T00:00Z" end="2009-12-31T00:00Z" timezone="America/Los_Angeles">
    <datasets>
    ...
    </datasets>
    ...
</cordinator-app>

Response for a bundle job:

HTTP/1.1 200 OK
Content-Type: application/xml;charset=UTF-8
.
<?xml version="1.0" encoding="UTF-8"?>
<bundle-app name='abc-app' xmlns="uri:oozie:coordinator:0.1" 
                 start="2009-01-01T00:00Z" end="2009-12-31T00:00Z"">
    <datasets>
    ...
    </datasets>
    ...
</bundle-app>

Job Log

An HTTP GET request retrieves the job log.

Request:

GET /oozie/v1/job/job-3?show=log

Response:

HTTP/1.1 200 OK
Content-Type: text/plain;charset=UTF-8
.
...
23:21:31,272 TRACE oozieapp:526 - USER[bansalm] GROUP[other] TOKEN[-] APP[test-wf] JOB[0-20090518232130-oozie-tucu] ACTION[mr-1] Start
23:21:31,305 TRACE oozieapp:526 - USER[bansalm] GROUP[other] TOKEN[-] APP[test-wf] JOB[0-20090518232130-oozie-tucu] ACTION[mr-1] End
...

Filtering the server logs with logfilter options

User can provide multiple option to filter logs using -logfilter opt1=val1;opt2=val1;opt3=val1. This can be used to fetch only just logs of interest faster as fetching Oozie server logs is slow due to the overhead of pattern matching.

GET /oozie/v1/job/0000003-140319184715726-oozie-puru-C?show=log&logfilter=limit=3;loglevel=WARN

Refer to the Filtering the server logs with logfilter options for more details.

Job graph

An HTTP GET request returns the image of the workflow DAG (rendered as a PNG image).

  • The nodes that are being executed are painted yellow
  • The nodes that have successfully executed are painted green
  • The nodes that have failed execution are painted red
  • The nodes that are yet to be executed are pained gray
  • An arc painted green marks the successful path taken so far
  • An arc painted red marks the failure of the node and highlights the error action
  • An arc painted gray marks a path not taken yet

Request:

GET /oozie/v1/job/job-3?show=graph[&show-kill=true]

Response:

HTTP/1.1 200 OK
Content-Type: image/png
Content-Length: {image_size_in_bytes}{image_bits}

The optional show-kill parameter shows kill node in the graph. Valid values for this parameter are 1 , yes , and true . This parameter has no effect when workflow fails and the failure node leads to the kill node; in that case kill node is shown always.

The node labels are the node names provided in the workflow XML.

This API returns HTTP 400 when run on a resource other than a workflow, viz. bundle and coordinator.

Jobs Information

A HTTP GET request retrieves workflow and coordinator jobs information.

Request:

GET /oozie/v1/jobs?filter=user%3Dbansalm&offset=1&len=50&timezone=GMT

Note that the filter is URL encoded, its decoded value is user=bansalm .

Response:

HTTP/1.1 200 OK
Content-Type: application/json;charset=UTF-8
.
{
  offset: 1,
  len: 50,
  total: 1002,
**jobs: [
    {
**    jobType: "workflow"
      id: "0-200905191240-oozie-W",
      appName: "indexer-workflow",
      appPath: "hdfs://user/tucu/indexer-wf",
      user: "bansalm",
      group: "other",
      status: "RUNNING",
      createdTime: "Thu, 01 Jan 2009 00:00:00 GMT",
      startTime: "Fri, 02 Jan 2009 00:00:00 GMT",
      endTime: null,
      info: "run=0",
    },
    {
**    jobType: "coordinator"
      id: "0-200905191240-oozie-C",
      appName: "logprocessor-coord",
      appPath: "hdfs://user/bansalm/myapp/logprocessor-coord.xml",
      user: "bansalm",
      group: "other",
      status: "RUNNING",
      createdTime: "Thu, 01 Jan 2009 00:00:00 GMT",
      startTime: "Fri, 02 Jan 2009 00:00:00 GMT",
      endTime: "Fri, 31 Dec 2009 00:00:00 GMT",
      info: "nextAction=5",
    },
    {
**    jobType: "bindle"
      id: "0-200905191240-oozie-B",
      appName: "logprocessor-bundle",
      appPath: "hdfs://user/bansalm/myapp/logprocessor-bundle.xml",
      user: "bansalm",
      group: "other",
      status: "RUNNING",
      createdTime: "Thu, 01 Jan 2009 00:00:00 GMT",
      startTime: "Fri, 02 Jan 2009 00:00:00 GMT",
      endTime: "Fri, 31 Dec 2009 00:00:00 GMT",
    },
    ...
  ]
}

No action information is returned when querying for multiple jobs.

The syntax for the filter is

[NAME=VALUE][;NAME=VALUE]*

Valid filter names are:

  • name: the application name from the workflow/coordinator/bundle definition
  • user: the user that submitted the job
  • group: the group for the job
  • status: the status of the job

The query will do an AND among all the filter names.

The query will do an OR among all the filter values for the same name. Multiple values must be specified as different name value pairs.

Additionally the offset and len parameters can be used for pagination. The start parameter is base 1.

Moreover, the jobtype parameter could be used to determine what type of job is looking for. The valid values of job type are: wf , coordinator or bundle .

Jobs information using Bulk API

A HTTP GET request retrieves a bulk response for all actions, corresponding to a particular bundle, that satisfy user specified criteria. This is useful for monitoring purposes, where user can find out about the status of downstream jobs with a single bulk request. The criteria are used for filtering the actions returned. Valid options (_case insensitive_) for these request criteria are:

  • bundle : the application name from the bundle definition
  • coordinators : the application name(s) from the coordinator definition.
  • actionStatus : the status of coordinator action (Valid values are WAITING, READY, SUBMITTED, RUNNING, SUSPENDED, TIMEDOUT, SUCCEEDED, KILLED, FAILED)
  • startCreatedTime : the start of the window you want to look at, of the actions' created time
  • endCreatedTime : the end of above window
  • startScheduledTime : the start of the window you want to look at, of the actions' scheduled i.e. nominal time.
  • endScheduledTime : the end of above window

Specifying 'bundle' is REQUIRED. All the rest are OPTIONAL but that might result in thousands of results depending on the size of your job. (pagination comes into play then)

If no 'actionStatus' values provided, by default KILLED,FAILED will be used. For e.g if the query string is only "bundle=MyBundle", the response will have all actions (across all coordinators) whose status is KILLED or FAILED

The query will do an AND among all the filter names, and OR among each filter name's values.

The syntax for the request criteria is

[NAME=VALUE][;NAME=VALUE]*

For 'coordinators' and 'actionStatus', if user wants to check for multiple values, they can be passed in a comma-separated manner. *Note*: The query will do an OR among them. Hence no need to repeat the criteria name

All the time values should be specified in ISO8601 (UTC) format i.e. yyyy-MM-dd'T'HH:mm'Z'

Additionally the offset and len parameters can be used as usual for pagination. The start parameter is base 1.

If you specify a coordinator in the list, that does not exist, no error is thrown; simply the response will be empty or pertaining to the other valid coordinators. However, if bundle name provided does not exist, an error is thrown.

Request:

GET /oozie/v1/jobs?bulk=bundle%3Dmy-bundle-app;coordinators%3Dmy-coord-1,my-coord-5;actionStatus%3DKILLED&offset=1&len=50

Note that the filter is URL encoded, its decoded value is user=chitnis . If typing in browser URL, one can type decoded value itself i.e. using '='

Response:

HTTP/1.1 200 OK
Content-Type: application/json;charset=UTF-8
.
{
  offset: 1,
  len: 50,
  total: 1002,
**  bulkresponses: [
**  {
      bulkbundle:
      {
        bundleJobName: "my-bundle-app",
        bundleJobId: "0-200905191240-oozie-B",
        status: "SUSPENDED",
      },
      bulkcoord:
      {
        coordJobName: "my-coord-1",
        status: "SUSPENDED",
      },
      bulkaction:
      {
        id: "0-200905191240-oozie-C@21",
        coordJobId: "0-200905191240-oozie-C",
        actionNumber: 21,
        externalId: "job_00076_0009",
        status: "KILLED",
        externalStatus: "FAILED",
        errorCode: "E0902",
        errorMessage: "Input file corrupt",
        createdTime: "Fri, 02 Jan 2009 00:00:00 GMT",
        nominalTime: "Thu, 01 Jan 2009 00:00:00 GMT",
        missingDependencies: "hdfs://nn:port/user/joe/file.txt"
      },
    },
**  {
      bulkbundle:
      {
        bundleJobName: "my-bundle-app",
        bundleJobId: "0-200905191240-oozie-B",
        status: "SUSPENDED",
      }
      bulkcoord:
      {
        coordJobName: "my-coord-5",
        status: "SUSPENDED",
      }
      bulkaction:
      {
        id: "0-200905191245-oozie-C@114",
        coordJobId: "0-200905191245-oozie-C",
        actionNumber: 114,
        externalId: "job_00076_0265",
        status: "KILLED",
        externalStatus: "KILLED",
        errorCode: "E0603",
        errorMessage: "SQL error in operation ...",
        createdTime: "Fri, 02 Jan 2009 00:00:00 GMT",
        nominalTime: "Thu, 01 Jan 2009 00:00:00 GMT",
        missingDependencies:
      }
    }
    ...
  ]
}

Oozie Web Services API, V2 (Workflow , Coordinator And Bundle)

The Oozie Web Services API is a HTTP REST JSON API.

All responses are in UTF-8 .

Assuming Oozie is runing at OOZIE_URL , the following web services end points are supported:

  • /versions
  • /v2/admin
  • /v2/job
  • /v2/jobs

Changes in v2 job API:

There is a difference in the JSON format of Job Information API (*/job) particularly for map-reduce action. No change for other actions. In v1, externalId and consoleUrl point to spawned child job ID, and exteranlChildIDs is null in map-reduce action. In v2, externalId and consoleUrl point to launcher job ID, and exteranlChildIDs is spawned child job ID in map-reduce action.

v2 supports retrieving of JMS topic on which job notifications are sent

REST API URL:

GET http://localhost:11000/oozie/v2/job/0000002-130507145349661-oozie-vira-W?show=jmstopic

Changes in v2 admin API:

v2 adds support for retrieving JMS connection information related to JMS notifications.

REST API URL:

GET http://localhost:11000/oozie/v2/admin/jmsinfo

v2/jobs remain the same as v1/jobs

Job and Jobs End-Points

Job Information

A HTTP GET request retrieves the job information.

Request:

GET /oozie/v2/job/job-3?show=info&timezone=GMT

Response for a workflow job:

HTTP/1.1 200 OK
Content-Type: application/json;charset=UTF-8
.
{
**jobType: "workflow",
  id: "0-200905191240-oozie-W",
  appName: "indexer-workflow",
  appPath: "hdfs://user/bansalm/indexer.wf",
  externalId: "0-200905191230-oozie-pepe",
  user: "bansalm",
  group: "other",
  status: "RUNNING",
  conf: "<configuration> ... </configuration>",
  createdTime: "Thu, 01 Jan 2009 00:00:00 GMT",
  startTime: "Fri, 02 Jan 2009 00:00:00 GMT",
  endTime: null,
  run: 0,
  actions: [
    {
      id: "0-200905191240-oozie-W@indexer",
      name: "indexer",
      type: "map-reduce",
      conf: "<configuration> ...</configuration>",
      startTime: "Thu, 01 Jan 2009 00:00:00 GMT",
      endTime: "Fri, 02 Jan 2009 00:00:00 GMT",
      status: "OK",
      externalId: "job-123-200903101010",
      externalStatus: "SUCCEEDED",
      trackerUri: "foo:8021",
      consoleUrl: "http://foo:50040/jobdetailshistory.jsp?jobId=job-123-200903101010",
      transition: "reporter",
      data: null,
      stats: null,
      externalChildIDs: "job-123-200903101011"
      errorCode: null,
      errorMessage: null,
      retries: 0
    },
    ...
  ]
}