Features

• Supports Spark Jobserver 0.6.0+

Library Installation

$ pip install python-sjsclient

Getting started

First create a client instance:

>>> from sjsclient import client
>>> sjs = client.Client("http://JOB_SERVER_URL:PORT")

Uploading a jar to Spark Jobserver:

>>> jar_file_path = os.path.join("path", "to", "jar")
>>> jar_blob = open(jar_file_path, 'rb').read()
>>> app = sjs.apps.create("test_app", jar_blob)

Uploading a python egg to Spark Jobserver:

>>> from sjsclient import app
>>> egg_file_path = os.path.join("path", "to", "egg")
>>> egg_blob = open(egg_file_path, 'rb').read()
>>> app = sjs.apps.create("test_python_app", egg_blob, app.AppType.PYTHON)

Listing available apps:

>>> for app in sjs.apps.list():
...     print app.name
...
test_app
my_streaming_app

Creating an adhoc job:

>>> test_app = sjs.apps.get("test_app")
>>> class_path = "spark.jobserver.VeryShortDoubleJob"
>>> config = {"test_config": "test_config_value"}
>>> job = sjs.jobs.create(test_app, class_path, conf=config)
>>> print("Job Status: ", job.status)
Job Status: STARTED

Creating a synchronous adhoc job:

>>> job = sjs.jobs.create(test_app, class_path, conf=config, sync=True)
>>> print(job.result)
[2, 4, 6]

Polling for job status:

>>> job = sjs.jobs.create(...)
>>> while job.status != "FINISHED":
>>>     time.sleep(2)
>>>     job = sjs.jobs.get(job.jobId)

Getting job config:

>>> config = {"test_config": "test_config_value"}
>>> job = sjs.jobs.create(test_app, class_path, conf=config)
>>> job_config = job.get_config()
>>> print("test_config value: ", job_config["test_config"])
test_config_value: test_config_value

Listing jobs:

>>> for job in sjs.jobs.list():
...     print job.jobId
...
8c5bd52f-6486-44ee-9ac3-a8327ee40494
24b67573-3115-49c7-983c-d0eff0499b71
99c8be9e-a0ec-42dd-8a2c-9a8680bc5051
bb82f712-d4b4-43a4-8e4d-e4bb272e85db

Limiting jobs list:

>>> for job in sjs.jobs.list(limit=1):
...     print job.jobId
...
8c5bd52f-6486-44ee-9ac3-a8327ee40494

Creating a named context:

>>> ctx_config = {'num-cpu-cores': '1', 'memory-per-node': '512m'}
>>> ctx = sjs.contexts.create("test_context", ctx_config)

Running a job in a named context:

>>> test_app = sjs.apps.get("test_app")
>>> test_ctx = sjs.contexts.get("test_context")
>>> config = {"test_config": "test_config_value"}
>>> job = sjs.jobs.create(test_app, class_path, ctx=test_ctx, conf=config)
>>> print("Job Status: ", job.status)
Job Status: STARTED

Documentation

http://python-sjsclient.readthedocs.org

Discussion list

spark-jobserver google group: https://groups.google.com/forum/#!forum/spark-jobserver

Requirements

• Python >= 2.7.0

License

python-sjsclient is offered under the Apache 2 license.

Source code

The latest developer version is available in a github repository: https://github.com/spark-jobserver/python-sjsclient