class AWSBstatCommand:
"""awsbstat command."""
__JOB_CONVERTERS = {"SIMPLE": JobConverter(), "ARRAY": ArrayJobConverter(), "MNP": MNPJobConverter()}
def __init__(self, log, boto3_factory):
"""
Initialize the object.
:param log: log
:param boto3_factory: an initialized Boto3ClientFactory object
"""
self.log = log
mapping = collections.OrderedDict(
[
("jobId", "id"),
("jobName", "name"),
("createdAt", "creation_time"),
("startedAt", "start_time"),
("stoppedAt", "stop_time"),
("status", "status"),
("statusReason", "status_reason"),
("jobDefinition", "job_definition"),
("jobQueue", "queue"),
("command", "command"),
("exitCode", "exit_code"),
("reason", "reason"),
("vcpus", "vcpus"),
("memory[MB]", "memory"),
("nodes", "nodes"),
("logStream", "log_stream"),
("log", "log_stream_url"),
("s3FolderUrl", "s3_folder_url"),
]
)
self.output = Output(mapping=mapping)
self.boto3_factory = boto3_factory
self.batch_client = boto3_factory.get_client("batch")
def run(self, job_status, expand_children, job_queue=None, job_ids=None, show_details=False):
"""Print list of jobs, by filtering by queue or by ids."""
if job_ids:
self.__populate_output_by_job_ids(job_ids, show_details or len(job_ids) == 1, include_parents=True)
# explicitly asking for job details,
# or asking for a single simple job (the output is not a list of jobs)
details_required = show_details or (len(job_ids) == 1 and self.output.length() == 1)
elif job_queue:
self.__populate_output_by_queue(job_queue, job_status, expand_children, show_details)
details_required = show_details
else:
fail("Error listing jobs from AWS Batch. job_ids or job_queue must be defined")
sort_keys_function = self.__sort_by_status_startedat_jobid() if not job_ids else self.__sort_by_key(job_ids)
if details_required:
self.output.show(sort_keys_function=sort_keys_function)
else:
self.output.show_table(
keys=["jobId", "jobName", "status", "startedAt", "stoppedAt", "exitCode"],
sort_keys_function=sort_keys_function,
)
@staticmethod
def __sort_by_key(ordered_keys): # noqa: D202
"""
Build a function to sort the output by key.
:param ordered_keys: list containing the sorted keys.
:return: a function to be used as key argument of the sorted function.
"""
def _sort_by_key(item):
job_id = item.id
try:
# in case the parent id was provided as input, sort children based on parent id position
parent_id = re.findall(r"[\w-]+", job_id)[0]
job_position = ordered_keys.index(parent_id)
except ValueError:
# in case the child id was provided as input, use its position in the list
job_position = ordered_keys.index(job_id)
return (
# sort by id according to the order in the keys_order list
job_position,
# sort by full id (needed to have parent before children)
job_id,
)
return _sort_by_key
@staticmethod
def __sort_by_status_startedat_jobid():
"""
Build a function to sort the output by (status, startedAt, jobId).
:return: a function to be used as key argument of the sorted function.
"""
return lambda item: (
# sort by status. Status order is defined by AWS_BATCH_JOB_STATUS.
AWS_BATCH_JOB_STATUS.index(item.status),
# sort by startedAt column.
item.start_time,
# sort by jobId column.
item.id,
)
def __populate_output_by_job_ids(self, job_ids, details, include_parents=False):
"""
Add Job item or jobs array children to the output.
:param job_ids: job ids or ARNs
:param details: ask for job details
"""
try:
if job_ids:
self.log.info("Describing jobs (%s), details (%s)" % (job_ids, details))
parent_jobs = []
jobs_with_children = []
jobs = self.__chunked_describe_jobs(job_ids)
for job in jobs:
# always add parent job
if include_parents or get_job_type(job) == "SIMPLE":
parent_jobs.append(job)
if is_job_array(job):
jobs_with_children.append((job["jobId"], ":", job["arrayProperties"]["size"]))
elif is_mnp_job(job):
jobs_with_children.append((job["jobId"], "#", job["nodeProperties"]["numNodes"]))
# add parent jobs to the output
self.__add_jobs(parent_jobs)
# create output items for jobs' children
self.__populate_output_by_parent_ids(jobs_with_children)
except Exception as e:
fail("Error describing jobs from AWS Batch. Failed with exception: %s" % e)
def __populate_output_by_parent_ids(self, parent_jobs):
"""
Add jobs children to the output.
:param parent_jobs: list of triplets (job_id, job_id_separator, job_size)
"""
try:
expanded_job_ids = []
for parent_job in parent_jobs:
expanded_job_ids.extend(
[
"{JOB_ID}{SEPARATOR}{INDEX}".format(JOB_ID=parent_job[0], SEPARATOR=parent_job[1], INDEX=i)
for i in range(0, parent_job[2])
]
)
if expanded_job_ids:
jobs = self.__chunked_describe_jobs(expanded_job_ids)
# forcing details to be False since already retrieved.
self.__add_jobs(jobs)
except Exception as e:
fail("Error listing job children. Failed with exception: %s" % e)
def __chunked_describe_jobs(self, job_ids):
"""
Submit calls to describe_jobs in batches of 100 elements each.
describe_jobs API call has a hard limit on the number of job that can be
retrieved with a single call. In case job_ids has more than 100 items, this function
distributes the describe_jobs call across multiple requests.
:param job_ids: list of ids for the jobs to describe.
:return: list of described jobs.
"""
jobs = []
for index in range(0, len(job_ids), 100):
jobs_chunk = job_ids[index : index + 100] # noqa: E203
jobs.extend(self.batch_client.describe_jobs(jobs=jobs_chunk)["jobs"])
return jobs
def __add_jobs(self, jobs, details=False):
"""
Get job info from AWS Batch and add to the output.
:param jobs: list of jobs items (output of the list_jobs function)
:param details: ask for job details
"""
try:
if jobs:
self.log.debug("Adding jobs to the output (%s)" % jobs)
if details:
self.log.info("Asking for jobs details")
jobs_to_show = self.__chunked_describe_jobs([job["jobId"] for job in jobs])
else:
jobs_to_show = jobs
for job in jobs_to_show:
self.log.debug("Adding job to the output (%s)", job)
job_converter = self.__JOB_CONVERTERS[get_job_type(job)]
self.output.add(job_converter.convert(job))
except KeyError as e:
fail("Error building Job item. Key (%s) not found." % e)
except Exception as e:
fail("Error adding jobs to the output. Failed with exception: %s" % e)
def __populate_output_by_queue(self, job_queue, job_status, expand_children, details):
"""
Add Job items to the output asking for given queue and status.
:param job_queue: job queue name or ARN
:param job_status: list of job status to ask
:param expand_children: if True, the job with children will be expanded by creating a row for each child
:param details: ask for job details
"""
try:
single_jobs = []
jobs_with_children = []
for status in job_status:
next_token = "" # nosec
while next_token is not None:
response = self.batch_client.list_jobs(jobStatus=status, jobQueue=job_queue, nextToken=next_token)
for job in response["jobSummaryList"]:
if get_job_type(job) != "SIMPLE" and expand_children is True:
jobs_with_children.append(job["jobId"])
else:
single_jobs.append(job)
next_token = response.get("nextToken")
# create output items for job array children
self.__populate_output_by_job_ids(jobs_with_children, details)
# add single jobs to the output
self.__add_jobs(single_jobs, details)
except Exception as e:
fail("Error listing jobs from AWS Batch. Failed with exception: %s" % e)
class AWSBstatCommand(object):
"""
awsbstat command
"""
def __init__(self, log, boto3_factory):
"""
:param log: log
:param boto3_factory: an initialized Boto3ClientFactory object
"""
self.log = log
mapping = collections.OrderedDict([('jobId', 'id'),
('jobName', 'name'),
('createdAt', 'creation_time'),
('startedAt', 'start_time'),
('stoppedAt', 'stop_time'),
('status', 'status'),
('statusReason', 'status_reason'),
('jobDefinition', 'job_definition'),
('jobQueue', 'queue'),
('command', 'command'),
('exitCode', 'exit_code'),
('reason', 'reason'),
('vcpus', 'vcpus'),
('memory[MB]', 'memory'),
('nodes', 'nodes'),
('logStream', 'log_stream'),
('log', 'log_stream_url')])
self.output = Output(mapping=mapping)
self.boto3_factory = boto3_factory
self.batch_client = boto3_factory.get_client('batch')
def run(self,
job_status,
expand_arrays,
job_queue=None,
job_ids=None,
show_details=False):
"""
print list of jobs, by filtering by queue or by ids
"""
if job_ids:
self.__populate_output_by_job_ids(
job_status, job_ids, show_details or len(job_ids) == 1)
# explicitly asking for job details,
# or asking for a single job that is not an array (the output is not a list of jobs)
details_required = show_details or (len(job_ids) == 1
and self.output.length() == 1)
elif job_queue:
self.__populate_output_by_queue(job_queue, job_status,
expand_arrays, show_details)
details_required = show_details
else:
fail(
"Error listing jobs from AWS Batch. job_ids or job_queue must be defined"
)
if details_required:
self.output.show()
else:
self.output.show_table([
'jobId', 'jobName', 'status', 'startedAt', 'stoppedAt',
'exitCode'
])
def __populate_output_by_job_ids(self, job_status, job_ids, details):
"""
Add Job item or jobs array children to the output
:param job_status: list of job status to ask
:param job_ids: job ids or ARNs
:param details: ask for job details
"""
try:
if job_ids:
self.log.info("Describing jobs (%s), details (%s)" %
(job_ids, details))
single_jobs = []
job_array_ids = []
jobs = self.batch_client.describe_jobs(jobs=job_ids)['jobs']
for job in jobs:
if is_job_array(job):
job_array_ids.append(job['jobId'])
else:
single_jobs.append(job)
# create output items for job array children
self.__populate_output_by_array_ids(job_status, job_array_ids,
details)
# add single jobs to the output
self.__add_jobs(single_jobs, details)
except Exception as e:
fail(
"Error describing jobs from AWS Batch. Failed with exception: %s"
% e)
def __populate_output_by_array_ids(self, job_status, job_array_ids,
details):
"""
Add jobs array children to the output
:param job_status: list of job status to ask
:param job_array_ids: job array ids to ask
:param details: ask for job details
"""
try:
for job_array_id in job_array_ids:
for status in job_status:
self.log.info(
"Listing job array children for job (%s) in status (%s)"
% (job_array_id, status))
next_token = ''
while next_token is not None:
response = self.batch_client.list_jobs(
jobStatus=status,
arrayJobId=job_array_id,
nextToken=next_token)
# add single jobs to the output
self.__add_jobs(response['jobSummaryList'], details)
next_token = response.get('nextToken')
except Exception as e:
fail(
"Error listing job array children for job (%s). Failed with exception: %s"
% (job_array_id, e))
def __add_jobs(self, jobs, details):
"""
Get job info from AWS Batch and add to the output
:param jobs: list of jobs items (output of the list_jobs function)
:param details: ask for job details
"""
try:
if jobs:
self.log.debug("Adding jobs to the output (%s)" % jobs)
if details:
self.log.info("Asking for jobs details")
jobs_to_show = []
for index in range(0, len(jobs), 100):
jobs_chunk = jobs[index:index + 100]
job_ids = []
for job in jobs_chunk:
job_ids.append(job['jobId'])
jobs_to_show.extend(
self.batch_client.describe_jobs(
jobs=job_ids)['jobs'])
else:
jobs_to_show = jobs
for job in jobs_to_show:
nodes = 1
if 'nodeProperties' in job:
# MNP job
container = job['nodeProperties'][
'nodeRangeProperties'][0]['container']
nodes = job['nodeProperties']['numNodes']
elif 'container' in job:
container = job['container']
else:
container = {}
if is_job_array(job):
# parent job array
job_id = '{0}[{1}]'.format(
job['jobId'], job['arrayProperties']['size'])
log_stream = '-'
log_stream_url = '-'
else:
job_id = job['jobId']
if 'logStreamName' in container:
log_stream = container.get('logStreamName')
log_stream_url = _compose_log_stream_url(
self.boto3_factory.region, log_stream)
else:
log_stream = '-'
log_stream_url = '-'
command = container.get('command', [])
self.log.debug("Adding job to the output (%s)", job)
job = Job(job_id=job_id,
name=job['jobName'],
creation_time=convert_to_date(job['createdAt']),
start_time=convert_to_date(job['startedAt'])
if 'startedAt' in job else '-',
stop_time=convert_to_date(job['stoppedAt'])
if 'stoppedAt' in job else '-',
status=job.get('status', 'UNKNOWN'),
status_reason=job.get('statusReason', '-'),
job_definition=get_job_definition_name_by_arn(
job['jobDefinition'], version=True)
if 'jobQueue' in job else '-',
queue=job['jobQueue'].split('/')[1]
if 'jobQueue' in job else '-',
command=shell_join(command) if command else '-',
reason=container.get('reason', '-'),
exit_code=container.get('exitCode', '-'),
vcpus=container.get('vcpus', '-'),
memory=container.get('memory', '-'),
nodes=nodes,
log_stream=log_stream,
log_stream_url=log_stream_url)
self.output.add(job)
except KeyError as e:
fail("Error building Job item. Key (%s) not found." % e)
except Exception as e:
fail("Error adding jobs to the output. Failed with exception: %s" %
e)
def __populate_output_by_queue(self, job_queue, job_status, expand_arrays,
details):
"""
Add Job items to the output asking for given queue and status
:param job_queue: job queue name or ARN
:param job_status: list of job status to ask
:param expand_arrays: if True, the job array will be expanded by creating a row for each child
:param details: ask for job details
"""
try:
for status in job_status:
next_token = ''
while next_token is not None:
response = self.batch_client.list_jobs(
jobStatus=status,
jobQueue=job_queue,
nextToken=next_token)
single_jobs = []
job_array_ids = []
for job in response['jobSummaryList']:
if is_job_array(job) and expand_arrays is True:
job_array_ids.append(job['jobId'])
else:
single_jobs.append(job)
# create output items for job array children
self.__populate_output_by_job_ids(job_status,
job_array_ids, details)
# add single jobs to the output
self.__add_jobs(single_jobs, details)
next_token = response.get('nextToken')
except Exception as e:
fail("Error listing jobs from AWS Batch. Failed with exception: %s"
% e)