def get(self, request, *args, **kwargs):
filepath = request.query_params.get('path')
if not filepath:
raise ValidationError('Files view expect a path to the file.')
job_outputs_path = get_job_outputs_path(
persistence_outputs=self.job.persistence_outputs,
job_name=self.job.unique_name)
download_filepath = archive_outputs_file(
persistence_outputs=self.job.persistence_outputs,
outputs_path=job_outputs_path,
namepath=self.job.unique_name,
filepath=filepath)
filename = os.path.basename(download_filepath)
chunk_size = 8192
try:
wrapped_file = FileWrapper(open(download_filepath, 'rb'),
chunk_size)
response = StreamingHttpResponse(
wrapped_file,
content_type=mimetypes.guess_type(download_filepath)[0])
response['Content-Length'] = os.path.getsize(download_filepath)
response['Content-Disposition'] = "attachment; filename={}".format(
filename)
return response
except FileNotFoundError:
_logger.warning('Log file not found: log_path=%s',
download_filepath)
return Response(status=status.HTTP_404_NOT_FOUND,
data='Log file not found: log_path={}'.format(
download_filepath))
def test_experiment_group_outputs_path_creation_deletion(self):
job = BuildJobFactory()
create_job_outputs_path(job.unique_name)
job_outputs_path = get_job_outputs_path(job.unique_name)
assert os.path.exists(job_outputs_path) is True
delete_job_outputs(job.unique_name)
assert os.path.exists(job_outputs_path) is False
def get_pod_container(self,
volume_mounts,
env_vars=None,
command=None,
args=None,
resources=None):
"""Pod job container for task."""
env_vars = get_list(env_vars)
env_vars += get_job_env_vars(
log_level=self.log_level,
outputs_path=get_job_outputs_path(job_name=self.job_name),
logs_path=get_job_logs_path(job_name=self.job_name),
data_path=get_job_data_path(job_name=self.job_name),
project_data_path=get_project_data_path(project_name=self.project_name)
)
env_vars += [
get_env_var(name=constants.CONFIG_MAP_JOB_INFO_KEY_NAME, value=json.dumps(self.labels)),
]
if resources:
env_vars += get_resources_env_vars(resources=resources)
ports = [client.V1ContainerPort(container_port=port) for port in self.ports]
return client.V1Container(name=self.job_container_name,
image=self.job_docker_image,
command=command,
args=args,
ports=ports,
env=env_vars,
resources=get_resources(resources),
volume_mounts=volume_mounts)
def get_pod_container(self,
volume_mounts,
env_vars=None,
command=None,
args=None,
resources=None):
"""Pod job container for task."""
env_vars = get_list(env_vars)
env_vars += get_job_env_vars(
log_level=self.log_level,
outputs_path=get_job_outputs_path(job_name=self.job_name),
logs_path=get_job_logs_path(job_name=self.job_name),
data_path=get_job_data_path(job_name=self.job_name),
project_data_path=get_project_data_path(project_name=self.project_name)
)
env_vars += [
get_env_var(name=constants.CONFIG_MAP_JOB_INFO_KEY_NAME, value=json.dumps(self.labels)),
]
if resources:
env_vars += get_resources_env_vars(resources=resources)
ports = [client.V1ContainerPort(container_port=port) for port in self.ports]
return client.V1Container(name=self.job_container_name,
image=self.job_docker_image,
command=command,
args=args,
ports=ports,
env=env_vars,
resources=get_resources(resources),
volume_mounts=volume_mounts)
def archive_job_outputs(persistence_outputs, job_name):
check_archive_path(settings.OUTPUTS_ARCHIVE_ROOT)
job_outputs_path = get_job_outputs_path(persistence_outputs=persistence_outputs,
job_name=job_name)
outputs_files = get_files_in_path(job_outputs_path)
tar_name = "{}.tar.gz".format(job_name.replace('.', '_'))
create_tarfile(files=outputs_files, tar_path=os.path.join(settings.OUTPUTS_ARCHIVE_ROOT,
tar_name))
return settings.OUTPUTS_ARCHIVE_ROOT, tar_name
def setUp(self):
super().setUp()
self.project = ProjectFactory(user=self.auth_client.user)
self.job = self.factory_class(project=self.project)
self.download_url = '/{}/{}/{}/jobs/{}/outputs'.format(
API_V1, self.project.user.username, self.project.name, self.job.id)
self.job_outputs_path = get_job_outputs_path(
persistence_outputs=self.job.persistence_outputs,
job_name=self.job.unique_name)
self.url = self.download_url
def get(self, request, *args, **kwargs):
auditor.record(event_type=JOB_OUTPUTS_DOWNLOADED,
instance=self.job,
actor_id=self.request.user.id,
actor_name=self.request.user.username)
job_outputs_path = get_job_outputs_path(
persistence_outputs=self.job.persistence_outputs,
job_name=self.job.unique_name)
archived_path, archive_name = archive_outputs(
outputs_path=job_outputs_path, name=self.job.unique_name)
return self.redirect(path='{}/{}'.format(archived_path, archive_name))
def test_job_outputs_path_creation_deletion(self):
job = JobFactory()
create_job_outputs_path(persistence_outputs=job.persistence_outputs,
job_name=job.unique_name)
job_outputs_path = get_job_outputs_path(
persistence_outputs=job.persistence_outputs,
job_name=job.unique_name)
assert os.path.exists(job_outputs_path) is True
delete_job_outputs(persistence_outputs=job.persistence_outputs,
job_name=job.unique_name)
assert os.path.exists(job_outputs_path) is False
def get_init_container(self, persistence_outputs):
"""Pod init container for setting outputs path."""
outputs_path = get_job_outputs_path(persistence_outputs=persistence_outputs,
job_name=self.job_name)
_, outputs_volume_mount = get_pod_outputs_volume(persistence_outputs=persistence_outputs)
return client.V1Container(
name=self.init_container_name,
image=self.init_docker_image,
command=["/bin/sh", "-c"],
args=to_list(get_output_args(command=InitCommands.CREATE,
outputs_path=outputs_path)),
volume_mounts=outputs_volume_mount)
def get_init_container(self):
"""Pod init container for setting outputs path."""
outputs_path = get_job_outputs_path(job_name=self.job_name)
outputs_volume_mount = get_volume_mount(
volume=constants.DATA_VOLUME,
volume_mount=settings.DATA_ROOT)
return client.V1Container(
name=self.init_container_name,
image=self.init_docker_image,
command=["/bin/sh", "-c"],
args=to_list(get_output_args(command=InitCommands.CREATE,
outputs_path=outputs_path)),
volume_mounts=[outputs_volume_mount])
def get_init_container(self):
"""Pod init container for setting outputs path."""
outputs_path = get_job_outputs_path(job_name=self.job_name)
outputs_volume_mount = get_volume_mount(
volume=constants.DATA_VOLUME, volume_mount=settings.DATA_ROOT)
return client.V1Container(name=self.init_container_name,
image=self.init_docker_image,
command=["/bin/sh", "-c"],
args=to_list(
get_output_args(
command=InitCommands.CREATE,
outputs_path=outputs_path)),
volume_mounts=[outputs_volume_mount])
def setUp(self):
super().setUp()
project = ProjectFactory(user=self.auth_client.user)
job = JobFactory(project=project)
self.url = '/{}/{}/{}/jobs/{}/outputs/files'.format(
API_V1, project.user.username, project.name, job.id)
outputs_path = get_job_outputs_path(
persistence_outputs=job.persistence_outputs,
job_name=job.unique_name)
create_job_outputs_path(persistence_outputs=job.persistence_outputs,
job_name=job.unique_name)
self.create_paths(path=outputs_path, url=self.url)
def get_pod_container(self,
volume_mounts,
persistence_outputs,
persistence_data,
outputs_refs_jobs=None,
outputs_refs_experiments=None,
secret_refs=None,
configmap_refs=None,
env_vars=None,
command=None,
args=None,
resources=None):
"""Pod job container for task."""
# Env vars preparation
env_vars = get_list(env_vars)
env_vars += get_job_env_vars(
log_level=self.log_level,
persistence_outputs=persistence_outputs,
outputs_path=get_job_outputs_path(
persistence_outputs=persistence_outputs,
job_name=self.job_name),
persistence_data=persistence_data,
logs_path=get_job_logs_path(job_name=self.job_name),
outputs_refs_jobs=outputs_refs_jobs,
outputs_refs_experiments=outputs_refs_experiments)
env_vars += [
get_env_var(name=constants.CONFIG_MAP_JOB_INFO_KEY_NAME,
value=json.dumps(self.labels)),
]
env_vars += get_resources_env_vars(resources=resources)
# Env from configmap and secret refs
env_from = get_pod_env_from(secret_refs=secret_refs,
configmap_refs=configmap_refs)
ports = [
client.V1ContainerPort(container_port=port) for port in self.ports
]
return client.V1Container(name=self.job_container_name,
image=self.job_docker_image,
command=command,
args=args,
ports=ports or None,
env=env_vars,
env_from=env_from,
resources=get_resources(resources),
volume_mounts=volume_mounts)
def get(self, request, *args, **kwargs):
store_manager = get_outputs_store(persistence_outputs=self.job.persistence_outputs)
job_outputs_path = get_job_outputs_path(
persistence_outputs=self.job.persistence_outputs,
job_name=self.job.unique_name)
if request.query_params.get('path'):
job_outputs_path = os.path.join(job_outputs_path,
request.query_params.get('path'))
try:
data = store_manager.ls(job_outputs_path)
except VolumeNotFoundError:
raise ValidationError('Store manager could not load the volume requested,'
' to get the outputs data.')
except Exception:
raise ValidationError('Experiment outputs path does not exists or bad configuration.')
return Response(data=data, status=200)
def get_jobs_outputs_spec(self):
from libs.paths.jobs import get_job_outputs_path
if not self.jobs.count():
return None
annotation = {
'persistence_outputs': KeyTransform('outputs', 'persistence')
}
query = self.jobs.annotate(**annotation)
job_data = query.values_list('id', 'project__user__username',
'project__name', 'persistence_outputs')
outputs_spec_data = {}
for data in job_data:
project_name = PROJECT_UNIQUE_NAME_FORMAT.format(user=data[1],
project=data[2])
job_name = JOB_UNIQUE_NAME_FORMAT.format(project_name=project_name,
id=data[0])
outputs_path = get_job_outputs_path(persistence_outputs=data[3],
job_name=job_name)
outputs_spec_data[data[0]] = OutputsRefsSpec(path=outputs_path,
persistence=data[3])
return outputs_spec_data
