def test_owner_and_label_switches(self):
runner_opts = ['--no-conf', '--owner=ads', '--label=ads_chain']
runner = MRTwoStepJob(runner_opts).make_runner()
match = JOB_KEY_RE.match(runner.get_job_key())
self.assertEqual(match.group(1), 'ads_chain')
self.assertEqual(match.group(2), 'ads')
def test_owner_and_label_kwargs(self):
runner = InlineMRJobRunner(conf_paths=[],
owner='ads', label='ads_chain')
match = JOB_KEY_RE.match(runner.get_job_key())
self.assertEqual(match.group(1), 'ads_chain')
self.assertEqual(match.group(2), 'ads')
def test_auto_owner(self):
os.environ['USER'] = '******'
runner = InlineMRJobRunner(conf_paths=[])
match = JOB_KEY_RE.match(runner.get_job_key())
self.assertEqual(match.group(1), 'no_script')
self.assertEqual(match.group(2), 'mcp')
def test_empty_no_user(self):
self.getuser_should_fail = True
runner = InlineMRJobRunner(conf_paths=[])
match = JOB_KEY_RE.match(runner.get_job_key())
self.assertEqual(match.group(1), 'no_script')
self.assertEqual(match.group(2), 'no_user')
def test_auto_everything(self):
test_start = datetime.datetime.utcnow()
os.environ['USER'] = '******'
runner = MRTwoStepJob(['--no-conf']).make_runner()
match = JOB_KEY_RE.match(runner.get_job_key())
self.assertEqual(match.group(1), 'mr_two_step_job')
self.assertEqual(match.group(2), 'mcp')
job_start = datetime.datetime.strptime(
match.group(3) + match.group(4), '%Y%m%d%H%M%S')
job_start = job_start.replace(microsecond=int(match.group(5)))
self.assertGreaterEqual(job_start, test_start)
self.assertLessEqual(job_start - test_start,
datetime.timedelta(seconds=5))
def cluster_to_basic_summary(cluster, now=None):
"""Extract fields such as creation time, owner, etc. from the job flow,
so we can safely reference them without using :py:func:`getattr`.
:param cluster: a :py:class:`boto.emr.EmrObject`
:param now: the current UTC time, as a :py:class:`datetime.datetime`.
Defaults to the current time.
Returns a dictionary with the following keys. These will be ``None`` if the
corresponding field in the job flow is unavailable.
* *created*: UTC `datetime.datetime` that the job flow was created,
or ``None``
* *end*: UTC `datetime.datetime` that the job flow finished, or ``None``
* *id*: job flow ID, or ``None`` (this should never happen)
* *label*: The label for the job flow (usually the module name of the
:py:class:`~mrjob.job.MRJob` script that started it), or
``None`` for non-:py:mod:`mrjob` job flows.
* *name*: job flow name, or ``None`` (this should never happen)
* *nih*: number of normalized instance hours used by the job flow.
* *num_steps*: Number of steps in the job flow.
* *owner*: The owner for the job flow (usually the user that started it),
or ``None`` for non-:py:mod:`mrjob` job flows.
* *pool*: pool name (e.g. ``'default'``) if the job flow is pooled,
otherwise ``None``.
* *ran*: How long the job flow ran, or has been running, as a
:py:class:`datetime.timedelta`. This will be ``timedelta(0)`` if
the job flow hasn't started.
* *ready*: UTC `datetime.datetime` that the job flow finished
bootstrapping, or ``None``
* *state*: The job flow's state as a string (e.g. ``'RUNNING'``)
"""
if now is None:
now = datetime.utcnow()
bcs = {} # basic cluster summary to fill in
bcs['id'] = getattr(cluster, 'id', None)
bcs['name'] = getattr(cluster, 'name', None)
status = getattr(cluster, 'status', None)
timeline = getattr(status, 'timeline', None)
bcs['created'] = to_datetime(getattr(
timeline, 'creationdatetime', None))
bcs['ready'] = to_datetime(getattr(timeline, 'readydatetime', None))
bcs['end'] = to_datetime(getattr(timeline, 'enddatetime', None))
if bcs['created']:
bcs['ran'] = (bcs['end'] or now) - bcs['created']
else:
bcs['ran'] = timedelta(0)
bcs['state'] = getattr(status, 'state', None)
bcs['num_steps'] = len(getattr(cluster, 'steps', ()))
bcs['pool'] = None
bootstrap_actions = getattr(cluster, 'bootstrapactions', None)
if bootstrap_actions:
args = [arg.value for arg in bootstrap_actions[-1].args]
if len(args) == 2 and args[0].startswith('pool-'):
bcs['pool'] = args[1]
m = JOB_KEY_RE.match(bcs['name'] or '')
if m:
bcs['label'], bcs['owner'] = m.group(1), m.group(2)
else:
bcs['label'], bcs['owner'] = None, None
bcs['nih'] = float(getattr(cluster, 'normalizedinstancehours', '0'))
return bcs
def cluster_to_basic_summary(cluster, now=None):
"""Extract fields such as creation time, owner, etc. from the cluster,
so we can safely reference them without using :py:func:`getattr`.
:param cluster: a :py:class:`boto.emr.EmrObject`
:param now: the current UTC time, as a :py:class:`datetime.datetime`.
Defaults to the current time.
Returns a dictionary with the following keys. These will be ``None`` if the
corresponding field in the cluster is unavailable.
* *created*: UTC `datetime.datetime` that the cluster was created,
or ``None``
* *end*: UTC `datetime.datetime` that the cluster finished, or ``None``
* *id*: cluster ID, or ``None`` (this should never happen)
* *label*: The label for the cluster (usually the module name of the
:py:class:`~mrjob.job.MRJob` script that started it), or
``None`` for non-:py:mod:`mrjob` clusters.
* *name*: cluster name, or ``None`` (this should never happen)
* *nih*: number of normalized instance hours used by the cluster.
* *num_steps*: Number of steps in the cluster.
* *owner*: The owner for the cluster (usually the user that started it),
or ``None`` for non-:py:mod:`mrjob` clusters.
* *pool*: pool name (e.g. ``'default'``) if the cluster is pooled,
otherwise ``None``.
* *ran*: How long the cluster ran, or has been running, as a
:py:class:`datetime.timedelta`. This will be ``timedelta(0)`` if
the cluster hasn't started.
* *ready*: UTC `datetime.datetime` that the cluster finished
bootstrapping, or ``None``
* *state*: The cluster's state as a string (e.g. ``'RUNNING'``)
"""
if now is None:
now = datetime.utcnow()
bcs = {} # basic cluster summary to fill in
bcs['id'] = getattr(cluster, 'id', None)
bcs['name'] = getattr(cluster, 'name', None)
status = getattr(cluster, 'status', None)
timeline = getattr(status, 'timeline', None)
bcs['created'] = to_datetime(getattr(
timeline, 'creationdatetime', None))
bcs['ready'] = to_datetime(getattr(timeline, 'readydatetime', None))
bcs['end'] = to_datetime(getattr(timeline, 'enddatetime', None))
if bcs['created']:
bcs['ran'] = (bcs['end'] or now) - bcs['created']
else:
bcs['ran'] = timedelta(0)
bcs['state'] = getattr(status, 'state', None)
bcs['num_steps'] = len(getattr(cluster, 'steps', ()))
bcs['pool'] = None
bootstrap_actions = getattr(cluster, 'bootstrapactions', None)
if bootstrap_actions:
args = [arg.value for arg in bootstrap_actions[-1].args]
if len(args) == 2 and args[0].startswith('pool-'):
bcs['pool'] = args[1]
m = JOB_KEY_RE.match(bcs['name'] or '')
if m:
bcs['label'], bcs['owner'] = m.group(1), m.group(2)
else:
bcs['label'], bcs['owner'] = None, None
bcs['nih'] = float(getattr(cluster, 'normalizedinstancehours', '0'))
return bcs
def test_auto_label(self):
runner = MRTwoStepJob(['--no-conf']).make_runner()
match = JOB_KEY_RE.match(runner.get_job_key())
self.assertEqual(match.group(1), 'mr_two_step_job')
self.assertEqual(match.group(2), getpass.getuser())
def test_empty(self):
runner = InlineMRJobRunner(conf_paths=[])
match = JOB_KEY_RE.match(runner.get_job_key())
self.assertEqual(match.group(1), 'no_script')
self.assertEqual(match.group(2), getpass.getuser())
