def ls(clusters, args, _):
"""Lists contents of the corresponding Mesos sandbox path by job or instance uuid."""
guard_no_cluster(clusters)
entity_refs, clusters_of_interest = parse_entity_refs(clusters, args.get('uuid'))
path = args.get('path')
long_format = args.get('long_format')
as_json = args.get('json')
literal = args.get('literal')
if len(entity_refs) > 1:
# argparse should prevent this, but we'll be defensive anyway
raise Exception(f'You can only provide a single uuid.')
if path and not literal and any(c in path for c in '*?[]{}'):
message = 'It looks like you are trying to glob, but ls does not support globbing. ' \
f'You can use the {terminal.bold("ssh")} command instead:\n' \
'\n' \
f' cs ssh {entity_refs[0]}\n' \
'\n' \
f'Or, if you want the literal path {terminal.bold(path)}, add {terminal.bold("--literal")}:\n' \
'\n' \
f' cs ls {terminal.bold("--literal")} {entity_refs[0]} {path}'
print(message)
return 1
command_fn = partial(ls_for_instance, path=path, long_format=long_format, as_json=as_json)
query_unique_and_run(clusters_of_interest, entity_refs[0], command_fn)
def show(clusters, args, _):
"""Prints info for the jobs / instances / groups with the given UUIDs."""
guard_no_cluster(clusters)
as_json = args.get('json')
entity_refs, _ = parse_entity_refs(clusters, args.get('uuid'))
query_result, clusters_of_interest = query_with_stdin_support(
clusters, entity_refs)
if as_json:
print(json.dumps(query_result))
else:
for cluster_name, entities in query_result['clusters'].items():
if 'jobs' in entities:
show_data(cluster_name, entities['jobs'], tabulate_job)
if 'instances' in entities:
show_data(cluster_name, entities['instances'],
tabulate_instance)
if 'groups' in entities:
show_data(cluster_name, entities['groups'], tabulate_group)
if query_result['count'] > 0:
return 0
else:
if not as_json:
print_no_data(clusters_of_interest)
return 1
def ssh(clusters, args, _):
"""Attempts to ssh (using os.execlp) to the Mesos agent corresponding to the given job or instance uuid."""
guard_no_cluster(clusters)
entity_refs, clusters_of_interest = parse_entity_refs(clusters, args.get('uuid'))
if len(entity_refs) > 1:
# argparse should prevent this, but we'll be defensive anyway
raise Exception(f'You can only provide a single uuid.')
query_unique_and_run(clusters_of_interest, entity_refs[0], ssh_to_instance)
def cat(clusters, args, _):
"""Outputs the contents of the corresponding Mesos sandbox path by job or instance uuid."""
guard_no_cluster(clusters)
entity_refs, clusters_of_interest = parse_entity_refs(clusters, args.get('target-entity'))
paths = args.get('path')
# argparse should prevent these, but we'll be defensive anyway
assert len(entity_refs) == 1, 'Only a single UUID or URL is supported.'
assert len(paths) == 1, 'Only a single path is supported.'
command_fn = partial(cat_for_instance, path=paths[0])
query_unique_and_run(clusters_of_interest, entity_refs[0], command_fn)
def kill(clusters, args, _):
"""Attempts to kill the jobs / instances / groups with the given UUIDs."""
guard_no_cluster(clusters)
entity_refs, _ = parse_entity_refs(clusters, args.get('uuid'))
query_result, clusters_of_interest = query_with_stdin_support(
clusters, entity_refs)
if query_result['count'] == 0:
print_no_data(clusters_of_interest)
return 1
# If the user provides UUIDs that map to more than one entity,
# we will raise an Exception that contains the details
guard_against_duplicates(query_result)
return kill_entities(query_result, clusters_of_interest)
def jobs(clusters, args, _):
"""Prints info for the jobs with the given list criteria"""
guard_no_cluster(clusters)
as_json = args.get('json')
one_per_line = args.get('one-per-line')
states = args.get('states')
user = args.get('user')
lookback_hours = args.get('lookback')
submitted_after = args.get('submitted_after')
submitted_before = args.get('submitted_before')
name = args.get('name')
limit = args.get('limit')
include_custom_executor = not args.get('exclude_custom_executor')
pool = args.get('pool')
if lookback_hours and (submitted_after or submitted_before):
raise Exception(
'You cannot specify both lookback hours and submitted after / before times.'
)
if submitted_after is not None or submitted_before is not None:
start_ms = date_time_string_to_ms_since_epoch(
submitted_after if submitted_after is not None else
f'{DEFAULT_LOOKBACK_HOURS} hours ago')
end_ms = date_time_string_to_ms_since_epoch(
submitted_before if submitted_before is not None else 'now')
else:
if states == ['running']:
default_lookback_hours = 24 * 7
else:
default_lookback_hours = DEFAULT_LOOKBACK_HOURS
start_ms, end_ms = lookback_hours_to_range(lookback_hours
or default_lookback_hours)
query_result = query(clusters, states, user, start_ms, end_ms, name, limit,
include_custom_executor, pool)
found_jobs = query_result['count'] > 0
if as_json:
print_as_json(query_result)
elif one_per_line:
print_as_one_per_line(query_result, clusters)
elif found_jobs:
print_as_table(query_result)
else:
print_no_data(clusters, states, user)
return 0
def wait(clusters, args, _):
"""Waits for jobs / instances / groups with the given UUIDs to complete."""
guard_no_cluster(clusters)
timeout = args.get('timeout')
interval = args.get('interval')
entity_refs, _ = parse_entity_refs(clusters, args.get('uuid'))
timeout_text = (
'up to %s' %
seconds_to_timedelta(timeout)) if timeout else 'indefinitely'
print_info('Will wait %s.' % timeout_text)
query_result, clusters_of_interest = query_with_stdin_support(
clusters, entity_refs, all_jobs_completed, all_instances_completed,
all_groups_completed, timeout, interval)
if query_result['count'] > 0:
return 0
else:
print_no_data(clusters_of_interest)
return 1
def usage(clusters, args, _):
"""Prints cluster usage info for the given user"""
guard_no_cluster(clusters)
as_json = args.get('json')
user = args.get('user')
pools = args.get('pool')
query_result = query(clusters, user)
if pools:
query_result = filter_query_result_by_pools(query_result, pools)
if as_json:
print_as_json(query_result)
else:
print_formatted(query_result)
return 0
def instances(clusters, args):
"""Prints the count of instances with the given criteria"""
guard_no_cluster(clusters)
if len(clusters) != 1:
raise Exception(f'You must specify a single cluster to query.')
status = args.get('status')
started_after = args.get('started_after')
started_before = args.get('started_before')
status = status or 'success'
start_ms = date_time_string_to_ms_since_epoch(started_after or '10 minutes ago')
end_ms = date_time_string_to_ms_since_epoch(started_before or 'now')
cluster = clusters[0]
data = query_instances_on_cluster(cluster, status, start_ms, end_ms)
print(data['count'])
return 0
def tail(clusters, args, _):
"""Tails the contents of the corresponding Mesos sandbox path by job or instance uuid."""
guard_no_cluster(clusters)
entity_refs, clusters_of_interest = parse_entity_refs(clusters, args.get('uuid'))
paths = args.get('path')
lines = args.get('lines')
follow = args.get('follow')
sleep_interval = args.get('sleep-interval')
if len(entity_refs) > 1:
# argparse should prevent this, but we'll be defensive anyway
raise Exception(f'You can only provide a single uuid.')
if len(paths) > 1:
# argparse should prevent this, but we'll be defensive anyway
raise Exception(f'You can only provide a single path.')
command_fn = partial(tail_for_instance, path=paths[0], num_lines_to_print=lines,
follow=follow, follow_sleep_seconds=sleep_interval)
query_unique_and_run(clusters_of_interest, entity_refs[0], command_fn)
def submit(clusters, args, _):
"""
Submits a job (or multiple jobs) to cook scheduler. Assembles a list of jobs,
potentially getting data from configuration, the command line, and stdin.
"""
guard_no_cluster(clusters)
logging.debug('submit args: %s' % args)
job_template = args
raw = job_template.pop('raw', None)
command_from_command_line = job_template.pop('command', None)
command_prefix = job_template.pop('command-prefix')
application_name = job_template.pop('application-name',
'cook-scheduler-cli')
application_version = job_template.pop('application-version',
version.VERSION)
job_template['application'] = {
'name': application_name,
'version': application_version
}
pool = job_template.pop('pool-name', None)
docker_image = job_template.pop('docker-image', None)
if docker_image:
job_template['container'] = {
'type': 'docker',
'docker': {
'image': docker_image,
'network': 'HOST',
'force-pull-image': False
}
}
group = None
if 'group-name' in job_template:
# If the user did not also specify a group uuid, generate
# one for them, and place the job(s) into the group
if 'group' not in job_template:
job_template['group'] = str(uuid.uuid4())
# The group name is specified on the group object
group = {
'name': job_template.pop('group-name'),
'uuid': job_template['group']
}
if raw:
if command_from_command_line:
raise Exception(
'You cannot specify a command at the command line when using --raw/-r.'
)
jobs_json = read_jobs_from_stdin()
jobs = parse_raw_job_spec(job_template, jobs_json)
else:
commands = acquire_commands(command_from_command_line)
if job_template.get('uuid') and len(commands) > 1:
raise Exception(
'You cannot specify multiple subcommands with a single UUID.')
if job_template.get('env'):
job_template['env'] = dict(
[e.split('=', maxsplit=1) for e in job_template['env']])
if job_template.get('label'):
labels = dict(
[l.split('=', maxsplit=1) for l in job_template['label']])
job_template.pop('label')
if 'labels' not in job_template:
job_template['labels'] = {}
job_template['labels'].update(labels)
jobs = [deep_merge(job_template, {'command': c}) for c in commands]
for job in jobs:
if not job.get('uuid'):
job['uuid'] = str(uuid.uuid4())
if not job.get('name'):
job['name'] = '%s_job' % current_user()
if command_prefix:
job['command'] = f'{command_prefix}{job["command"]}'
logging.debug('jobs: %s' % jobs)
return submit_federated(clusters, jobs, group, pool)
def submit(clusters, args, _):
"""
Submits a job (or multiple jobs) to cook scheduler. Assembles a list of jobs,
potentially getting data from configuration, the command line, and stdin.
"""
guard_no_cluster(clusters)
logging.debug('submit args: %s' % args)
job_template = args
raw = job_template.pop('raw', None)
command_from_command_line = job_template.pop('command', None)
command_prefix = job_template.pop('command-prefix')
application_name = job_template.pop('application-name',
'cook-scheduler-cli')
application_version = job_template.pop('application-version',
version.VERSION)
job_template['application'] = {
'name': application_name,
'version': application_version
}
pool = job_template.pop('pool-name', None)
checkpoint = job_template.pop('checkpoint', False)
checkpoint_mode = job_template.pop('checkpoint-mode', None)
checkpoint_preserve_paths = job_template.pop('checkpoint-preserve-paths',
None)
checkpoint_period_sec = job_template.pop('checkpoint-period-sec', None)
disk_request = job_template.pop('disk-request', None)
disk_limit = job_template.pop('disk-limit', None)
disk_type = job_template.pop('disk-type', None)
docker_image = job_template.pop('docker-image', None)
if docker_image:
job_template['container'] = {
'type': 'docker',
'docker': {
'image': docker_image,
'network': 'HOST',
'force-pull-image': False
}
}
group = None
if 'group-name' in job_template:
# If the user did not also specify a group uuid, generate
# one for them, and place the job(s) into the group
if 'group' not in job_template:
job_template['group'] = str(make_temporal_uuid())
# The group name is specified on the group object
group = {
'name': job_template.pop('group-name'),
'uuid': job_template['group']
}
if raw:
if command_from_command_line:
raise Exception(
'You cannot specify a command at the command line when using --raw/-r.'
)
jobs_json = read_jobs_from_stdin()
jobs = parse_raw_job_spec(job_template, jobs_json)
else:
commands = acquire_commands(command_from_command_line)
if job_template.get('uuid') and len(commands) > 1:
raise Exception(
'You cannot specify multiple subcommands with a single UUID.')
if job_template.get('env'):
job_template['env'] = dict(
[e.split('=', maxsplit=1) for e in job_template['env']])
if job_template.get('label'):
labels = dict(
[l.split('=', maxsplit=1) for l in job_template['label']])
job_template.pop('label')
if 'labels' not in job_template:
job_template['labels'] = {}
job_template['labels'].update(labels)
if job_template.get('constraint'):
constraints = []
for c in job_template['constraint']:
parts = c.split('=', maxsplit=1)
if len(parts) != 2:
raise Exception(
'Invalid constraint, must be of the form K=V')
constraints.append([parts[0], 'EQUALS', parts[1]])
job_template.pop('constraint')
job_template['constraints'] = constraints
jobs = [deep_merge(job_template, {'command': c}) for c in commands]
for job in jobs:
if not job.get('uuid'):
job['uuid'] = str(make_temporal_uuid())
if not job.get('name'):
job['name'] = '%s_job' % current_user()
if command_prefix:
job['command'] = f'{command_prefix}{job["command"]}'
if checkpoint or checkpoint_mode:
checkpoint = {
'mode': checkpoint_mode if checkpoint_mode else 'auto'
}
if checkpoint_preserve_paths:
checkpoint['options'] = {
'preserve-paths': checkpoint_preserve_paths
}
if checkpoint_period_sec:
checkpoint['periodic-options'] = {
'period-sec': checkpoint_period_sec
}
job['checkpoint'] = checkpoint
if disk_request or disk_limit or disk_type:
disk = {}
if disk_request:
disk['request'] = disk_request
if disk_limit:
disk['limit'] = disk_limit
if disk_type:
disk['type'] = disk_type
job['disk'] = disk
logging.debug('jobs: %s' % jobs)
return submit_federated(clusters, jobs, group, pool)
