def test_add_steps_to_flow():
conn = boto.connect_emr()
step1 = StreamingStep(
name='My wordcount example',
mapper='s3n://elasticmapreduce/samples/wordcount/wordSplitter.py',
reducer='aggregate',
input='s3n://elasticmapreduce/samples/wordcount/input',
output='s3n://output_bucket/output/wordcount_output')
job_id = conn.run_jobflow(name='My jobflow',
log_uri='s3://some_bucket/jobflow_logs',
steps=[step1])
job_flow = conn.describe_jobflow(job_id)
job_flow.state.should.equal('STARTING')
job_flow.jobflowid.should.equal(job_id)
job_flow.name.should.equal('My jobflow')
job_flow.loguri.should.equal('s3://some_bucket/jobflow_logs')
step2 = StreamingStep(
name='My wordcount example2',
mapper='s3n://elasticmapreduce/samples/wordcount/wordSplitter2.py',
reducer='aggregate',
input='s3n://elasticmapreduce/samples/wordcount/input2',
output='s3n://output_bucket/output/wordcount_output2')
conn.add_jobflow_steps(job_id, [step2])
job_flow = conn.describe_jobflow(job_id)
job_step = job_flow.steps[0]
job_step.name.should.equal('My wordcount example')
job_step.state.should.equal('STARTING')
args = [arg.value for arg in job_step.args]
args.should.equal([
'-mapper',
's3n://elasticmapreduce/samples/wordcount/wordSplitter.py',
'-reducer',
'aggregate',
'-input',
's3n://elasticmapreduce/samples/wordcount/input',
'-output',
's3n://output_bucket/output/wordcount_output',
])
job_step2 = job_flow.steps[1]
job_step2.name.should.equal('My wordcount example2')
job_step2.state.should.equal('PENDING')
args = [arg.value for arg in job_step2.args]
args.should.equal([
'-mapper',
's3n://elasticmapreduce/samples/wordcount/wordSplitter2.py',
'-reducer',
'aggregate',
'-input',
's3n://elasticmapreduce/samples/wordcount/input2',
'-output',
's3n://output_bucket/output/wordcount_output2',
])
def add_step_emr(conn, cluster_id):
step = StreamingStep(name='MC_Method example',
cache_files=['s3n://bucket774/map.py#map.py'],
mapper='map.py',
input='s3://bucket774/input/',
output='s3://bucket774/output/')
conn.add_jobflow_steps(cluster_id, step)
def test_create_instance_groups():
conn = boto.connect_emr()
step1 = StreamingStep(
name='My wordcount example',
mapper='s3n://elasticmapreduce/samples/wordcount/wordSplitter.py',
reducer='aggregate',
input='s3n://elasticmapreduce/samples/wordcount/input',
output='s3n://output_bucket/output/wordcount_output')
job_id = conn.run_jobflow(
name='My jobflow',
log_uri='s3://some_bucket/jobflow_logs',
steps=[step1],
)
instance_group = InstanceGroup(6, 'TASK', 'c1.medium', 'SPOT', 'spot-0.07',
'0.07')
instance_group = conn.add_instance_groups(job_id, [instance_group])
instance_group_id = instance_group.instancegroupids
job_flow = conn.describe_jobflows()[0]
int(job_flow.instancecount).should.equal(6)
instance_group = job_flow.instancegroups[0]
instance_group.instancegroupid.should.equal(instance_group_id)
int(instance_group.instancerunningcount).should.equal(6)
instance_group.instancerole.should.equal('TASK')
instance_group.instancetype.should.equal('c1.medium')
instance_group.market.should.equal('SPOT')
instance_group.name.should.equal('spot-0.07')
instance_group.bidprice.should.equal('0.07')
def create_emr(R):
if not boto.config.has_section('Boto'):
boto.config.add_section('Boto')
boto.config.set('Boto', 'https_validate_certificates', 'False')
step = StreamingStep(name='MC_Method example',
cache_files=['s3n://bucket774/map.py#map.py'],
mapper='map.py',
input='s3://bucket774/input/',
output='s3://bucket774/output/')
conn = EmrConnection(access_id, access_key)
instance_groups = []
instance_groups.append(
InstanceGroup(num_instances=1,
role="MASTER",
type='m4.large',
market="ON_DEMAND",
name="Master nodes"))
if R > 1:
instance_groups.append(
InstanceGroup(num_instances=R - 1,
role="CORE",
type='m4.large',
market="ON_DEMAND",
name="Slave nodes"))
cluster_id = conn.run_jobflow(name='test MC_method run',
instance_groups=instance_groups,
enable_debugging=False,
steps=[step],
visible_to_all_users=True,
keep_alive=True,
job_flow_role="EMR_EC2_DefaultRole",
service_role="EMR_DefaultRole",
hadoop_version='2.4.0',
log_uri='s3://bucket774/log')
return cluster_id, conn
def _make_step(self,
mapper,
reducer,
input,
output,
num_mappers=1,
num_reducers=1):
"""
Returns a new step that runs the specified mapper and reducer,
reading from the specified input and writing to the specified
output.
"""
bucket = self._s3_conn.get_bucket(self._s3_bucket)
# Clear out current bucket/output contents for team
keys = bucket.list(prefix=self._get_keyname(output))
bucket.delete_keys(keys)
step_name = self._make_name()
step_args = [
'-jobconf',
'mapred.map.tasks=%d' % (num_mappers), '-jobconf',
'mapred.reduce.tasks=%d' % (num_reducers)
]
return StreamingStep(name=step_name,
step_args=step_args,
mapper=self._get_s3_team_uri(mapper),
reducer=self._get_s3_team_uri(reducer),
input=self._get_s3_team_uri(input),
output=self._get_s3_team_uri(output))
def setup_and_run_job(self):
""" Runs the Elastic MapReduce job on AWS"""
step = StreamingStep(
name='Titanic Machine Learning',
mapper='s3n://' + EmrProcessing.bucket_name + '/mapper/mapper.py',
reducer='org.apache.hadoop.mapred.lib.IdentityReducer',
input='s3n://' + EmrProcessing.bucket_name + '/input/',
output='s3n://' + EmrProcessing.bucket_name + '/output/')
self.conn = connect_to_region(self.region_name)
self.jobid = self.conn.run_jobflow(
name='Titanic Devp',
log_uri='s3://' + EmrProcessing.bucket_name + '/jobflow_logs',
steps=[step])
def createNewStreamingStep(stepName, fileList, outputDirectory, mapperKey,
mapperLocation, reducerLocation):
if reducerLocation != 'aggregate':
args = ['-files', mapperLocation + ',' + reducerLocation]
reducerKey = reducerLocation.split('/')[-1]
else:
args = ['-files', mapperLocation]
reducerKey = reducerLocation
return StreamingStep(name=stepName,
mapper=mapperKey,
reducer=reducerKey,
input=fileList,
output=outputDirectory,
step_args=args)
def _make_step(self, mapper, reducer, input, output, nm=1, nr=1):
job_name = self._make_name()
team_s3 = self._get_s3_url()
bucket = self.s3_conn.get_bucket(self.s3_bucket)
keys = bucket.list(prefix='%s/%s' % (self.team_id, output))
bucket.delete_keys(map(lambda k: k.name, keys))
return \
StreamingStep(name=job_name,
step_args=
['-jobconf', 'mapred.map.tasks=%d' % nm,
'-jobconf', 'mapred.reduce.tasks=%d' % nr],
mapper=team_s3 + mapper,
reducer=team_s3 + reducer,
input=team_s3 + input,
output=team_s3 + output)
def _create_sa_job(emr_conn):
"""Create and start an EMR job asynchronously"""
step = StreamingStep(name='reTOracle Sentiment Analysis Step',
mapper='s3n://retoracle/sa_mapper.py',
reducer='s3n://retoracle/sa_reducer.py',
input='s3n://retoracle/sa_input',
output='s3n://retoracle/sa_output')
return emr_conn.run_jobflow(
name='reTOracle Sentiment Analysis Job',
log_uri='s3://retoracle/jobflow_logs',
steps=[step],
num_instances=EMR_NUM_TOTAL_NODES,
master_instance_type=EMR_TYPE_MASTER_NODE,
slave_instance_type=EMR_TYPE_SLAVE_NODES,
ec2_keyname=EMR_KP,
enable_debugging=EMR_DEBUGGING,
ami_version=EMR_AMI_VERSION,
)
def test_modify_instance_groups():
conn = boto.connect_emr()
step1 = StreamingStep(
name='My wordcount example',
mapper='s3n://elasticmapreduce/samples/wordcount/wordSplitter.py',
reducer='aggregate',
input='s3n://elasticmapreduce/samples/wordcount/input',
output='s3n://output_bucket/output/wordcount_output')
job_id = conn.run_jobflow(name='My jobflow',
log_uri='s3://some_bucket/jobflow_logs',
steps=[step1])
instance_group1 = InstanceGroup(6, 'TASK', 'c1.medium', 'SPOT',
'spot-0.07', '0.07')
instance_group2 = InstanceGroup(6, 'TASK', 'c1.medium', 'SPOT',
'spot-0.07', '0.07')
instance_group = conn.add_instance_groups(
job_id, [instance_group1, instance_group2])
instance_group_ids = instance_group.instancegroupids.split(",")
job_flow = conn.describe_jobflows()[0]
int(job_flow.instancecount).should.equal(12)
instance_group = job_flow.instancegroups[0]
int(instance_group.instancerunningcount).should.equal(6)
conn.modify_instance_groups(instance_group_ids, [2, 3])
job_flow = conn.describe_jobflows()[0]
int(job_flow.instancecount).should.equal(5)
instance_group1 = [
group for group in job_flow.instancegroups
if group.instancegroupid == instance_group_ids[0]
][0]
int(instance_group1.instancerunningcount).should.equal(2)
instance_group2 = [
group for group in job_flow.instancegroups
if group.instancegroupid == instance_group_ids[1]
][0]
int(instance_group2.instancerunningcount).should.equal(3)
def run_streaming_step(self, cluster_id, name, mapper_path, reducer_path, input_path, output_path):
try:
# bundle files with the job
files = []
if mapper_path != "NONE":
files.append(mapper_path)
mapper_path = mapper_path.split("/")[-1]
if reducer_path != "NONE":
files.append(reducer_path)
reducer_path = reducer_path.split("/")[-1]
# build streaming step
logging.debug("Launching streaming step with mapper: " + mapper_path + " reducer: " + reducer_path + " and files: " + str(files))
step = StreamingStep(name=name,
step_args=["-files"] + files,
mapper=mapper_path,
reducer=reducer_path,
input=input_path,
output=output_path,
action_on_failure="CONTINUE")
return self._run_step(cluster_id, step)
except:
logging.error("Running streaming step in cluster " + cluster_id + " failed.")
return "FAILED"
import logging
from boto.emr.step import StreamingStep
from emr_manager import EmrManager
# script to trigger a test step process
logging.getLogger().setLevel(logging.INFO)
manager = EmrManager()
step = StreamingStep(name="testStep", mapper="s3n://tuiinnovation-holycrap/scripts/mapper.py", reducer="NONE", input="s3n://tuiinnovation-emr/input/SuppliersMonitor.log-20140524.bz2", output="s3n://tuiinnovation-emr/output/", action_on_failure="CONTINUE")
cluster_id = "j-1OSKIJWRXGEJW"
step_id = manager.run_step(cluster_id, step)
logging.info("step " + step_id + " completed in " + cluster_id)
print word
# <codecell>
### Running code with EMR
#emrcon = EmrConnection('<aws access key>', '<aws secret key>')
emrcon = EmrConnection('AKIAJRV3RN6NXQTSSTBA',
'3e212d6rs99xtiPgwKnfN1QD30WZk2hJwCWjMcGc')
# <codecell>
# Using EMR's wordcount example
step = StreamingStep(
name='My wordcount example',
mapper='s3n://elasticmapreduce/samples/wordcount/wordSplitter.py',
reducer='aggregate',
input='s3n://elasticmapreduce/samples/wordcount/input',
output='s3n://wambia660fall2013/output/wordcount_output')
# <codecell>
jobid = emrcon.run_jobflow(name='Word Count Example',
log_uri='s3://wambia660fall2013/logs',
steps=[step])
# <codecell>
print jobid
# <codecell>
def test_create_job_flow():
conn = boto.connect_emr()
step1 = StreamingStep(
name='My wordcount example',
mapper='s3n://elasticmapreduce/samples/wordcount/wordSplitter.py',
reducer='aggregate',
input='s3n://elasticmapreduce/samples/wordcount/input',
output='s3n://output_bucket/output/wordcount_output')
step2 = StreamingStep(
name='My wordcount example2',
mapper='s3n://elasticmapreduce/samples/wordcount/wordSplitter2.py',
reducer='aggregate',
input='s3n://elasticmapreduce/samples/wordcount/input2',
output='s3n://output_bucket/output/wordcount_output2')
job_id = conn.run_jobflow(
name='My jobflow',
log_uri='s3://some_bucket/jobflow_logs',
master_instance_type='m1.medium',
slave_instance_type='m1.small',
steps=[step1, step2],
)
job_flow = conn.describe_jobflow(job_id)
job_flow.state.should.equal('STARTING')
job_flow.jobflowid.should.equal(job_id)
job_flow.name.should.equal('My jobflow')
job_flow.masterinstancetype.should.equal('m1.medium')
job_flow.slaveinstancetype.should.equal('m1.small')
job_flow.loguri.should.equal('s3://some_bucket/jobflow_logs')
job_flow.visibletoallusers.should.equal('False')
int(job_flow.normalizedinstancehours).should.equal(0)
job_step = job_flow.steps[0]
job_step.name.should.equal('My wordcount example')
job_step.state.should.equal('STARTING')
args = [arg.value for arg in job_step.args]
args.should.equal([
'-mapper',
's3n://elasticmapreduce/samples/wordcount/wordSplitter.py',
'-reducer',
'aggregate',
'-input',
's3n://elasticmapreduce/samples/wordcount/input',
'-output',
's3n://output_bucket/output/wordcount_output',
])
job_step2 = job_flow.steps[1]
job_step2.name.should.equal('My wordcount example2')
job_step2.state.should.equal('PENDING')
args = [arg.value for arg in job_step2.args]
args.should.equal([
'-mapper',
's3n://elasticmapreduce/samples/wordcount/wordSplitter2.py',
'-reducer',
'aggregate',
'-input',
's3n://elasticmapreduce/samples/wordcount/input2',
'-output',
's3n://output_bucket/output/wordcount_output2',
])
def test_steps():
input_steps = [
StreamingStep(
name="My wordcount example",
mapper="s3n://elasticmapreduce/samples/wordcount/wordSplitter.py",
reducer="aggregate",
input="s3n://elasticmapreduce/samples/wordcount/input",
output="s3n://output_bucket/output/wordcount_output",
),
StreamingStep(
name="My wordcount example & co.",
mapper="s3n://elasticmapreduce/samples/wordcount/wordSplitter2.py",
reducer="aggregate",
input="s3n://elasticmapreduce/samples/wordcount/input2",
output="s3n://output_bucket/output/wordcount_output2",
),
]
# TODO: implementation and test for cancel_steps
conn = boto.connect_emr()
cluster_id = conn.run_jobflow(steps=[input_steps[0]], **run_jobflow_args)
jf = conn.describe_jobflow(cluster_id)
jf.steps.should.have.length_of(1)
conn.add_jobflow_steps(cluster_id, [input_steps[1]])
jf = conn.describe_jobflow(cluster_id)
jf.steps.should.have.length_of(2)
for step in jf.steps:
step.actiononfailure.should.equal("TERMINATE_JOB_FLOW")
list(arg.value for arg in step.args).should.have.length_of(8)
step.creationdatetime.should.be.a(str)
# step.enddatetime.should.be.a(str)
step.jar.should.equal(
"/home/hadoop/contrib/streaming/hadoop-streaming.jar")
step.laststatechangereason.should.be.a(str)
step.mainclass.should.equal("")
step.name.should.be.a(str)
# step.readydatetime.should.be.a(str)
# step.startdatetime.should.be.a(str)
step.state.should.be.within(["RUNNING", "PENDING"])
expected = dict((s.name, s) for s in input_steps)
steps = conn.list_steps(cluster_id).steps
for x in steps:
y = expected[x.name]
# actiononfailure
list(arg.value for arg in x.config.args).should.equal([
"-mapper",
y.mapper,
"-reducer",
y.reducer,
"-input",
y.input,
"-output",
y.output,
])
x.config.jar.should.equal(
"/home/hadoop/contrib/streaming/hadoop-streaming.jar")
x.config.mainclass.should.equal("")
# properties
x.should.have.property("id").should.be.a(str)
x.name.should.equal(y.name)
x.status.state.should.be.within(["RUNNING", "PENDING"])
# x.status.statechangereason
x.status.timeline.creationdatetime.should.be.a(str)
# x.status.timeline.enddatetime.should.be.a(str)
# x.status.timeline.startdatetime.should.be.a(str)
x = conn.describe_step(cluster_id, x.id)
list(arg.value for arg in x.config.args).should.equal([
"-mapper",
y.mapper,
"-reducer",
y.reducer,
"-input",
y.input,
"-output",
y.output,
])
x.config.jar.should.equal(
"/home/hadoop/contrib/streaming/hadoop-streaming.jar")
x.config.mainclass.should.equal("")
# properties
x.should.have.property("id").should.be.a(str)
x.name.should.equal(y.name)
x.status.state.should.be.within(["RUNNING", "PENDING"])
# x.status.statechangereason
x.status.timeline.creationdatetime.should.be.a(str)
# x.status.timeline.enddatetime.should.be.a(str)
# x.status.timeline.startdatetime.should.be.a(str)
@requires_boto_gte("2.39")
def test_list_steps_with_states():
# boto's list_steps prior to 2.39 has a bug that ignores
# step_states argument.
steps = conn.list_steps(cluster_id).steps
step_id = steps[0].id
steps = conn.list_steps(cluster_id, step_states=["RUNNING"]).steps
steps.should.have.length_of(1)
steps[0].id.should.equal(step_id)
test_list_steps_with_states()
def test_steps():
input_steps = [
StreamingStep(
name='My wordcount example',
mapper='s3n://elasticmapreduce/samples/wordcount/wordSplitter.py',
reducer='aggregate',
input='s3n://elasticmapreduce/samples/wordcount/input',
output='s3n://output_bucket/output/wordcount_output'),
StreamingStep(
name='My wordcount example2',
mapper='s3n://elasticmapreduce/samples/wordcount/wordSplitter2.py',
reducer='aggregate',
input='s3n://elasticmapreduce/samples/wordcount/input2',
output='s3n://output_bucket/output/wordcount_output2')
]
# TODO: implementation and test for cancel_steps
conn = boto.connect_emr()
cluster_id = conn.run_jobflow(
steps=[input_steps[0]],
**run_jobflow_args)
jf = conn.describe_jobflow(cluster_id)
jf.steps.should.have.length_of(1)
conn.add_jobflow_steps(cluster_id, [input_steps[1]])
jf = conn.describe_jobflow(cluster_id)
jf.steps.should.have.length_of(2)
for step in jf.steps:
step.actiononfailure.should.equal('TERMINATE_JOB_FLOW')
list(arg.value for arg in step.args).should.have.length_of(8)
step.creationdatetime.should.be.a(six.string_types)
# step.enddatetime.should.be.a(six.string_types)
step.jar.should.equal(
'/home/hadoop/contrib/streaming/hadoop-streaming.jar')
step.laststatechangereason.should.be.a(six.string_types)
step.mainclass.should.equal('')
step.name.should.be.a(six.string_types)
# step.readydatetime.should.be.a(six.string_types)
# step.startdatetime.should.be.a(six.string_types)
step.state.should.be.within(['STARTING', 'PENDING'])
expected = dict((s.name, s) for s in input_steps)
steps = conn.list_steps(cluster_id).steps
for x in steps:
y = expected[x.name]
# actiononfailure
list(arg.value for arg in x.config.args).should.equal([
'-mapper', y.mapper,
'-reducer', y.reducer,
'-input', y.input,
'-output', y.output,
])
x.config.jar.should.equal(
'/home/hadoop/contrib/streaming/hadoop-streaming.jar')
x.config.mainclass.should.equal('')
# properties
x.should.have.property('id').should.be.a(six.string_types)
x.name.should.equal(y.name)
x.status.state.should.be.within(['STARTING', 'PENDING'])
# x.status.statechangereason
x.status.timeline.creationdatetime.should.be.a(six.string_types)
# x.status.timeline.enddatetime.should.be.a(six.string_types)
# x.status.timeline.startdatetime.should.be.a(six.string_types)
x = conn.describe_step(cluster_id, x.id)
list(arg.value for arg in x.config.args).should.equal([
'-mapper', y.mapper,
'-reducer', y.reducer,
'-input', y.input,
'-output', y.output,
])
x.config.jar.should.equal(
'/home/hadoop/contrib/streaming/hadoop-streaming.jar')
x.config.mainclass.should.equal('')
# properties
x.should.have.property('id').should.be.a(six.string_types)
x.name.should.equal(y.name)
x.status.state.should.be.within(['STARTING', 'PENDING'])
# x.status.statechangereason
x.status.timeline.creationdatetime.should.be.a(six.string_types)
# x.status.timeline.enddatetime.should.be.a(six.string_types)
# x.status.timeline.startdatetime.should.be.a(six.string_types)
@requires_boto_gte('2.39')
def test_list_steps_with_states():
# boto's list_steps prior to 2.39 has a bug that ignores
# step_states argument.
steps = conn.list_steps(cluster_id).steps
step_id = steps[0].id
steps = conn.list_steps(cluster_id, step_states=['STARTING']).steps
steps.should.have.length_of(1)
steps[0].id.should.equal(step_id)
test_list_steps_with_states()
from boto.emr.connection import EmrConnection
from boto.emr.step import StreamingStep
import boto
AWS_KEY='AKIAIQ7VG4UORIN75ZSA'
AWS_SECRET='jzxajGx8gzwX+ymYXJ0/5heCjkPtWLQkICYRn7Vj'
conn = EmrConnection(AWS_KEY, AWS_SECRET)
step = StreamingStep(name='My wordcount example',
mapper='s3n://css739/wordcount/bigramSplitter.py',
reducer='aggregate',
input='s3n://smalldata/wikipedia_titles.txt',
output='s3n://css739/wordcount/bigram_count_output2',
cache_files=['s3n://css739/wordcount/english_stoplist.py'])
jobid = conn.run_jobflow(name='My jobflow', log_uri='s3n://css739/wordcount/jobflow_logs',steps=[step])
conn.describe_jobflow(jobid).state
