def create_workflow_object():
"""
Creates the diamond DAG from the workflow examples
Looks something like this where we flow from left to right
one -> (two, three) -> four
"""
container = client.V1Container(name="busybox",
image="gcr.io/google-containers/busybox",
command=[
"sh", "-c",
"echo Starting on: $(date); sleep 5; \
echo Goodbye cruel world at: $(date)"
])
stepone = DagV1WorkflowStep(
name="stepone",
job_template=client.V1beta1JobTemplateSpec(spec=client.V1JobSpec(
template=client.V1PodTemplateSpec(
metadata=client.V1ObjectMeta(labels={"workflow": "stepone"}),
spec=client.V1PodSpec(containers=[container],
restart_policy="Never")))))
steptwo = DagV1WorkflowStep(
name="steptwo",
dependencies=['stepone'],
job_template=client.V1beta1JobTemplateSpec(spec=client.V1JobSpec(
template=client.V1PodTemplateSpec(
metadata=client.V1ObjectMeta(labels={"workflow": "steptwo"}),
spec=client.V1PodSpec(containers=[container],
restart_policy="Never")))))
stepthree = DagV1WorkflowStep(
name="stepthree",
dependencies=['stepone'],
job_template=client.V1beta1JobTemplateSpec(spec=client.V1JobSpec(
template=client.V1PodTemplateSpec(
metadata=client.V1ObjectMeta(labels={"workflow": "stepthree"}),
spec=client.V1PodSpec(containers=[container],
restart_policy="Never")))))
stepfour = DagV1WorkflowStep(
name="stepfour",
dependencies=['steptwo', 'stepthree'],
job_template=client.V1beta1JobTemplateSpec(spec=client.V1JobSpec(
template=client.V1PodTemplateSpec(
metadata=client.V1ObjectMeta(labels={"workflow": "stepfour"}),
spec=client.V1PodSpec(containers=[container],
restart_policy="Never")))))
workflow = DagV1Workflow(
metadata=client.V1ObjectMeta(name="diamond"),
spec=DagV1WorkflowSpec(selector=client.V1LabelSelector(
match_labels={"workflow": "diamond"}),
steps=[stepone, steptwo, stepthree, stepfour]))
return workflow
def create_k8s_job(self, job: KubernetesCMDJob) -> client.V1Job:
volume_mount = client.V1VolumeMount(name='download-volume', mount_path=DEFAULT_PROJECT_PATH)
image = job.job_config.properties.get('ai_flow_worker_image')
if isinstance(job.exec_cmd, List):
cmd = job.exec_cmd
else:
cmd = [job.exec_cmd]
working_dir = KubernetesJobPlugin.get_container_working_dir(job)
job_container = client.V1Container(name='cmd-job',
image=image,
image_pull_policy='Always',
command=cmd,
working_dir=working_dir,
volume_mounts=[volume_mount])
pod_spec = KubernetesJobPlugin.create_init_container(job, volume_mount, job_container)
labels = {'app': 'ai-flow', 'component': 'cmd-job-' + str(job.instance_id)}
object_meta = client.V1ObjectMeta(labels=labels,
annotations={ANNOTATION_WATCHED: 'True',
ANNOTATION_JOB_ID: str(job.instance_id),
ANNOTATION_JOB_UUID: str(job.uuid),
ANNOTATION_WORKFLOW_ID: str(
job.job_context.workflow_execution_id)})
template_spec = client.V1PodTemplateSpec(metadata=object_meta,
spec=pod_spec)
job_spec = client.V1JobSpec(template=template_spec, backoff_limit=0)
object_meta = client.V1ObjectMeta(labels=labels,
name=self.generate_job_name(job))
job = client.V1Job(metadata=object_meta, spec=job_spec, api_version='batch/v1', kind='Job')
return job
def create_job_manifest(envs, commands, name, image, template_file):
if template_file is not None:
with open(template_file) as f:
job = yaml.safe_load(f)
job["metadata"]["name"] = name
job["spec"]["template"]["metadata"]["labels"]["app"] = name
job["spec"]["template"]["spec"]["containers"][0]["image"] = image
job["spec"]["template"]["spec"]["containers"][0][
"command"] = commands
job["spec"]["template"]["spec"]["containers"][0]["name"] = name
job["spec"]["template"]["spec"]["containers"][0]["env"] = envs
job["spec"]["template"]["spec"]["containers"][0][
"command"] = commands
else:
container = client.V1Container(image=image,
command=commands,
name=name,
env=envs)
pod_temp = client.V1PodTemplateSpec(
spec=client.V1PodSpec(restart_policy="OnFailure",
containers=[container]),
metadata=client.V1ObjectMeta(name=name, labels={"app": name}))
job = client.V1Job(api_version="batch/v1",
kind="Job",
spec=client.V1JobSpec(template=pod_temp),
metadata=client.V1ObjectMeta(name=name))
return job
def create_job_object():
# Configureate Pod template container
# volume inside of which you put your service account
# This is the secret name, basically
volume_name = "google-cloud-json"
google_app_credentials_path = os.environ.get(
'GOOGLE_APPLICATION_CREDENTIALS')
# create a volume mount
volume_mount = client.V1VolumeMount(mount_path='/etc/stuff',
name=volume_name)
# Create environment variables for container.
# In this case, grab the values from the execution environment
# perhaps using something like a .env file.
env = [
client.V1EnvVar(
name='GOOGLE_APPLICATION_CREDENTIALS',
# note this is a path + the filename of the app creds in the secret
value='/etc/stuff/key.json'), # google_app_credentials_path),
client.V1EnvVar(name='GCS_STAGING_URL',
value=os.environ.get('GCS_STAGING_URL')),
client.V1EnvVar(name='GCS_EXPORT_URL',
value=os.environ.get('GCS_EXPORT_URL')),
client.V1EnvVar(name='ES_HOST', value=os.environ.get('ES_HOST'))
]
# Create a volume.
# This will go into the spec section of the template
# Note that this specifies a secret volume
# The secret needs to be created separately.
volume = client.V1Volume(
name=volume_name,
secret=client.V1SecretVolumeSource(secret_name='google-cloud-json'))
# Create the container section that will
# go into the spec section
container = client.V1Container(
name="omicidx-builder",
image="seandavi/omicidx-builder",
volume_mounts=[volume_mount],
env=env,
command=['/bin/bash', '/code/biosample_pipeline.sh'])
# Create and configurate a spec section
template = client.V1PodTemplateSpec(
metadata=client.V1ObjectMeta(labels={"app": "omicidx-builder"}),
spec=client.V1PodSpec(restart_policy="Never",
volumes=[volume],
containers=[container]))
# Create the specification of deployment
spec = client.V1JobSpec(template=template, backoff_limit=4)
# Instantiate the job object
job = client.V1Job(api_version="batch/v1",
kind="Job",
metadata=client.V1ObjectMeta(name=JOB_NAME),
spec=spec)
return job
def create_job_object(
self,
job_name,
container_image,
args,
envs,
resources,
label_key,
label_value,
backoff_limit=Constants.BACKOFF_LIMIT,
):
try:
pod_template_spec = self.create_pod_template_spec(
job_name, container_image, args, envs, resources)
# Create and configure a spec section
template = client.V1PodTemplateSpec(
metadata=client.V1ObjectMeta(labels={label_key: label_value}),
spec=pod_template_spec)
# Create the specification of deployment
spec = client.V1JobSpec(template=template,
backoff_limit=backoff_limit,
ttl_seconds_after_finished=3 * 24 * 60 *
60)
# Instantiate the job object
job = client.V1Job(api_version="batch/v1",
kind="Job",
metadata=client.V1ObjectMeta(name=job_name),
spec=spec)
return job
except:
print('error create_job_object')
def create(self, job_spec=None, pod_spec=None):
"""Creates and runs the job on the cluster.
Args:
job_spec: A dictionary of keyword arguments that will be passed to V1JobSpec()
pod_spec: A dictionary of keyword arguments that will be passed to V1PodSpec()
Returns: A dictionary containing the results of creating the job on the cluster.
"""
if job_spec is None:
job_spec = self.job_spec
if pod_spec is None:
pod_spec = self.pod_spec
if not self._containers:
raise ValueError(
"Containers not found. "
"Use add_containers() to specify containers before creating the job."
)
# TODO: Set the backoff limit to 1. There will be no retry if the job fails.
# Convert job name to lower case
job_name = str(self.job_name).lower()
job_body = client.V1Job(kind="Job")
job_body.metadata = client.V1ObjectMeta(namespace=self.namespace, name=job_name)
job_body.status = client.V1JobStatus()
template = pod_template(self._containers, self._volumes, **pod_spec)
job_body.spec = client.V1JobSpec(template=template.template, **job_spec)
self.creation_response = api_request(api.create_namespaced_job, self.namespace, job_body)
return self.creation_response
def generate_deployment_spec(self, pod_template_spec):
"""Generate a V1Job initialized with correct completion and
parallelism (for HP search) and with the provided V1PodTemplateSpec
:param pod_template_spec: V1PodTemplateSpec
"""
if not isinstance(pod_template_spec, k8s_client.V1PodTemplateSpec):
raise TypeError("""pod_template_spec must be a V1PodTemplateSpec,
but got %s""" % type(pod_template_spec))
job_spec = k8s_client.V1JobSpec(
template=pod_template_spec,
parallelism=self.runs,
completions=self.runs,
backoff_limit=0,
)
return k8s_client.V1Job(api_version="batch/v1",
kind="Job",
metadata=k8s_client.V1ObjectMeta(
name=self.job_name,
generate_name=constants.JOB_DEFAULT_NAME,
labels=self.labels),
spec=job_spec)
def make_job_object(filename,filepath,id_training):
env_vars=None
if env_vars is None:
env_vars ={"ENDPOINT":ENDPOINT ,"MINIO_ACCESS_KEY":MINIO_ACCESS_KEY , "MINIO_SECRET_KEY" : MINIO_SECRET_KEY , "BUCKET": BUCKET , "FILENAME":str(filename), "FILEPATH": str(filepath),"ID_TRAINING":str(id_training)}
env_list = []
for env_name, env_value in env_vars.items():
env_list.append( client.V1EnvVar(name=env_name, value=env_value) )
container = client.V1Container(
name="estratto",
image="ziofededocker/dummy:latest",
command=["python","/app/estrazione.py"],
env=env_list,
image_pull_policy="IfNotPresent")
template = client.V1PodTemplateSpec(
metadata=client.V1ObjectMeta(labels={"app": "estratto"}),
spec=client.V1PodSpec(restart_policy="Never", containers=[container]))
spec = client.V1JobSpec(
template=template,
backoff_limit=0)
job = client.V1Job(
api_version="batch/v1",
kind="Job",
metadata=client.V1ObjectMeta(name="estrazione-job"),
spec=spec)
return job
def create_job_object(self, job_name, container_image, args):
volume_name = "" # volume inside of which you put your service account
google_app_credentials_path = os.environ.get(
'GOOGLE_APPLICATION_CREDENTIALS')
volume_mount = client.V1VolumeMount(mount_path='/'.join(
google_app_credentials_path.split('/')[:-1]),
name=volume_name)
env = client.V1EnvVar(name='GOOGLE_APPLICATION_CREDENTIALS',
value=google_app_credentials_path)
container = client.V1Container(name=job_name,
image=container_image,
args=args,
volume_mounts=[volume_mount],
env=[env],
image_pull_policy="Always")
volume = client.V1Volume(
name=volume_name,
secret=client.V1SecretVolumeSource(
secret_name='<secret-where-you-put-the-service-account>'))
template = client.V1PodTemplateSpec(
metadata=client.V1ObjectMeta(labels={"app": "sample"}),
spec=client.V1PodSpec(restart_policy="Never",
containers=[container],
volumes=[volume]))
spec = client.V1JobSpec(template=template,
backoff_limit=3,
ttl_seconds_after_finished=60)
job = client.V1Job(api_version="batch/v1",
kind="Job",
metadata=client.V1ObjectMeta(name=job_name),
spec=spec)
return job
def create_job_object(message, environment_image):
"""Function to create the AWS EKS Job object
Arguments:
message {[dict]} -- Submission message from AWS SQS queue
Returns:
[AWS EKS Job class object] -- AWS EKS Job class object
"""
PYTHONUNBUFFERED_ENV = client.V1EnvVar(name="PYTHONUNBUFFERED", value="1")
AUTH_TOKEN_ENV = client.V1EnvVar(name="AUTH_TOKEN", value=AUTH_TOKEN)
EVALAI_API_SERVER_ENV = client.V1EnvVar(
name="EVALAI_API_SERVER", value=EVALAI_API_SERVER
)
MESSAGE_BODY_ENV = client.V1EnvVar(name="BODY", value=json.dumps(message))
submission_pk = message["submission_pk"]
image = message["submitted_image_uri"]
# Configureate Pod agent container
agent_container = client.V1Container(
name="agent", image=image, env=[PYTHONUNBUFFERED_ENV]
)
volume_mount_list = get_volume_mount_list("/dataset")
# Configureate Pod environment container
environment_container = client.V1Container(
name="environment",
image=environment_image,
env=[
PYTHONUNBUFFERED_ENV,
AUTH_TOKEN_ENV,
EVALAI_API_SERVER_ENV,
MESSAGE_BODY_ENV,
],
resources=client.V1ResourceRequirements(
limits={"nvidia.com/gpu": "1"}
),
volume_mounts=volume_mount_list,
)
volume_list = get_volume_list()
# Create and configurate a spec section
template = client.V1PodTemplateSpec(
metadata=client.V1ObjectMeta(labels={"app": "evaluation"}),
spec=client.V1PodSpec(
containers=[environment_container, agent_container],
restart_policy="Never",
volumes=volume_list,
),
)
# Create the specification of deployment
spec = client.V1JobSpec(backoff_limit=1, template=template)
# Instantiate the job object
job = client.V1Job(
api_version="batch/v1",
kind="Job",
metadata=client.V1ObjectMeta(
name="submission-{0}".format(submission_pk)
),
spec=spec,
)
return job
def create_job_object(env_vars={"ENDPOINT":"http://172.17.0.1:9000","MINIO_ACCESS_KEY":"admin","MINIO_SECRET_KEY":"keystone"}):
env_list = []
for env_name, env_value in env_vars.items():
env_list.append( client.V1EnvVar(name=env_name, value=env_value) )
container = client.V1Container(
name="estratto",
image="ziofededocker/estrazionefile:latest",
command=["python","/app/estrazione.py"],
env=env_list,
image_pull_policy="IfNotPresent")
template = client.V1PodTemplateSpec(
metadata=client.V1ObjectMeta(labels={"app": "estratto"}),
spec=client.V1PodSpec(restart_policy="Never", containers=[container]))
spec = client.V1JobSpec(
template=template,
backoff_limit=0)
job = client.V1Job(
api_version="batch/v1",
kind="Job",
metadata=client.V1ObjectMeta(name=JOB_NAME),
spec=spec)
return job
def create_job_object(container_image, image_pull_secret=None,service_account_name=None):
pull_secret = client.V1LocalObjectReference(
name=image_pull_secret
)
# Configureate Pod template container
container = client.V1Container(
name="installer",
image=container_image,
command=list(install_command()))
# Create and configurate a spec section
template = client.V1PodTemplateSpec(
metadata=client.V1ObjectMeta(labels={"app": "mcm-installer"}),
spec=client.V1PodSpec(restart_policy="Never", containers=[container],
image_pull_secrets=[pull_secret],service_account_name=service_account_name))
# Create the specification of deployment
spec = client.V1JobSpec(
template=template,
backoff_limit=1)
# Instantiate the job object
job = client.V1Job(
api_version="batch/v1",
kind="Job",
metadata=client.V1ObjectMeta(name=JOB_NAME),
spec=spec)
return job
def create_cronjob(self, schedule: str, pipeline_uuid: str) -> str:
run_job_endpoint= f'{kalytical_config.kalytical_api_endpoint}/pipeline/dispatcher/run_by_pipeline_uuid?pipeline_uuid={pipeline_uuid}'
job_name = f'kalytical-api-trigger-{pipeline_uuid}'
container = client.V1Container(
name=job_name,
image=kalytical_config.ext_cron_image_uri,
env=[client.V1EnvVar(name='KALYTICAL_API_ENDPOINT', value=run_job_endpoint),
client.V1EnvVar(name='KALYTICAL_API_AUTH_SECRET', value=kalytical_config.kalytical_api_token)],
resources=client.V1ResourceRequirements(limits={'cpu':'.1', 'memory':'50Mi'}))
pod_template = client.V1PodTemplateSpec(
metadata=client.V1ObjectMeta(labels={'kalytical-api-pipeline': job_name}),
spec=client.V1PodSpec(restart_policy="Never", containers=[container]))
job_spec = client.V1JobSpec(
completions=1, backoff_limit=0, template=pod_template)
job_template = client.V1beta1JobTemplateSpec(job_tepmlate = job_template, schedule=schedule),
cron_body = client.V1beta1CronJob(
spec=cron_spec, metadata=client.V1ObjectMeta(name=job_name)
)
try:
self.log.debug(f"Attempting to write namespaced cronjob with namespace={self._k8s_namespace} parameters={str(cron_body)}")
self._k8s_batch_client.create_namespaced_cron_job(
namespace=self._k8s_namespace, body=cron_body)
except ApiException as e:
if e.status == 409:
self.log.warn("This job already existed. We will re-create it.")
self.delete_cronjob(job_name=job_name) #TODO Instead use patching
self.create_cronjob(schedule=schedule, pipeline_uuid=pipeline_uuid)
else:
raise e
return job_name
def create_cronjob(username, namespace, dbhost):
try:
config.load_kube_config()
except:
config.load_incluster_config()
api = client.BatchV1beta1Api()
body = client.V1beta1CronJob(
metadata=client.V1ObjectMeta(name=namespace),
spec=client.V1beta1CronJobSpec( job_template=client.V1beta1JobTemplateSpec(
spec=client.V1JobSpec(template=client.V1PodTemplateSpec(
spec=client.V1PodSpec(
containers=[
client.V1Container(name="scheduler", image="sahandha/lsstscheduler",
args=["/bin/bash","-c","python /sched.py {} {} {};".format(username, namespace, dbhost)],
resources=client.V1ResourceRequirements(
requests={'memory': "200Mi", 'cpu': "100m"})
)],
restart_policy="OnFailure"
)))
),
schedule = "*/1 * * * *")
)
try:
api = api.create_namespaced_cron_job(namespace, body)
except ApiException as e:
print("Exception when calling BatchV1beta1Api->create_namespaced_cron_job: %s\n" % e)
def create_job(self, name, image, cmd, path):
container = client.V1Container(
name=name,
image=image,
env=[client.V1EnvVar(name='PYTHONUNBUFFERED', value='0')],
command=cmd,
volume_mounts=[
client.V1VolumeMount(
name=name + "-volume",
mount_path="/root",
)
])
volume = client.V1Volume(name=name + "-volume",
host_path=client.V1HostPathVolumeSource(
path=path, ))
template = client.V1PodTemplateSpec(metadata=client.V1ObjectMeta(
name=name, labels={"user": self.user}),
spec=client.V1PodSpec(
restart_policy="Never",
containers=[container],
volumes=[volume],
))
spec = client.V1JobSpec(template=template)
job = client.V1Job(api_version="batch/v1",
kind="Job",
metadata=client.V1ObjectMeta(name=name),
spec=spec)
client.BatchV1Api().create_namespaced_job(namespace=self.namespace,
body=job)
def create_job_object(runner_image, region, s3_path, pvc_name):
target_folder = get_target_folder(s3_path)
# Configureate Pod template container
container = k8s_client.V1Container(
name="copy-dataset-worker",
image=runner_image,
command=["aws"],
args=["s3", "sync", s3_path, "/mnt/" + target_folder],
volume_mounts=[k8s_client.V1VolumeMount(name="data-storage", mount_path='/mnt')],
env=[k8s_client.V1EnvVar(name="AWS_REGION", value=region),
k8s_client.V1EnvVar(name="AWS_ACCESS_KEY_ID", value_from=k8s_client.V1EnvVarSource(secret_key_ref=k8s_client.V1SecretKeySelector(key="AWS_ACCESS_KEY_ID", name="aws-secret"))),
k8s_client.V1EnvVar(name="AWS_SECRET_ACCESS_KEY", value_from=k8s_client.V1EnvVarSource(secret_key_ref=k8s_client.V1SecretKeySelector(key="AWS_SECRET_ACCESS_KEY", name="aws-secret")))
],
)
volume = k8s_client.V1Volume(
name='data-storage',
persistent_volume_claim=k8s_client.V1PersistentVolumeClaimVolumeSource(claim_name=pvc_name)
)
# Create and configurate a spec section
template = k8s_client.V1PodTemplateSpec(
# metadata=k8s_client.V1ObjectMeta(labels={"app":"copy-dataset-worker"}),
spec=k8s_client.V1PodSpec(containers=[container], volumes=[volume], restart_policy="OnFailure"))
# Create the specification of deployment
spec = k8s_client.V1JobSpec(
# selector=k8s_client.V1LabelSelector(match_labels={"app":"copy-dataset-worker"}),
template=template)
# Instantiate the deployment object
deployment = k8s_client.V1Job(
api_version="batch/v1",
kind="Job",
metadata=k8s_client.V1ObjectMeta(name=container.name),
spec=spec)
return deployment
def generate_app_migrator_job(self, tag: str, source: str):
log.debug("Generating app-migrator job: tag={} source={}".format(tag, source))
deployment = self.appsV1Api.read_namespaced_deployment(source, self.namespace)
metadata = client.V1ObjectMeta(
labels={"app": APP_MIGRATOR}, name=APP_MIGRATOR, namespace=self.namespace
)
new_image = generate_image(
old_image=deployment.spec.template.spec.containers[0].image, new_tag=tag
)
job = client.V1Job(
api_version="batch/v1",
kind="Job",
metadata=metadata,
spec=client.V1JobSpec(
template=client.V1PodTemplateSpec(
spec=deployment.spec.template.spec, metadata=metadata
)
),
)
job.spec.template.spec.containers[0].image = new_image
job.spec.template.spec.restart_policy = "Never"
job.spec.template.spec.containers[0].command = config.APP_MIGRATOR_COMMAND
job.spec.template.spec.containers[0].args = config.APP_MIGRATOR_ARGS
job.spec.template.spec.containers[0].resources = client.V1ResourceRequirements()
self.batchV1Api.create_namespaced_job(self.namespace, job)
log.debug(
"Generation of app-migrator job complete: tag={} source={}".format(
tag, source
)
)
def create_job_manifest(n_comp, n_para):
# 引数で渡された情報にもとづき, Kubernetes の job を生成する
container = client.V1Container(
name="job-worker",
image="masato0921/tutorial-docker-kubernetes:chapter_010_006",
env=[
client.V1EnvVar(name="BROKER_URL",
value="amqp://*****:*****@task-queue:5672"),
client.V1EnvVar(name="QUEUE", value="task-queue")
])
template = client.V1PodTemplateSpec(
spec=client.V1PodSpec(containers=[container], restart_policy="Never"))
spec = client.V1JobSpec(backoff_limit=4,
template=template,
completions=n_comp,
parallelism=n_para)
job = client.V1Job(api_version="batch/v1",
kind="Job",
metadata=client.V1ObjectMeta(name=OBJECT_NAME),
spec=spec)
return job
def getBody(namespace='couture-console',
jobname='nifi-test',
containername='nifi-test',
containerimage='sidharthc/nifi-test:alpha',
env_vars=ENV_LIST,
containerargs=['SFTP_TO_HDFS.py']):
body = client.V1Job(api_version="batch/v1", kind="Job")
# Body needs Metadata
# Attention: Each JOB must have a different name!
body.metadata = client.V1ObjectMeta(namespace=namespace, name=jobname)
# And a Status
body.status = client.V1JobStatus()
# Now we start with the Template...
template = client.V1PodTemplate()
template.template = client.V1PodTemplateSpec()
env_list = []
for env_name, env_value in env_vars.items():
env_list.append(client.V1EnvVar(name=env_name, value=env_value))
container = client.V1Container(name=containername,
image=containerimage,
args=containerargs,
env=env_list)
template.template.spec = client.V1PodSpec(containers=[container],
restart_policy='Never')
# And finaly we can create our V1JobSpec!
body.spec = client.V1JobSpec(ttl_seconds_after_finished=100,
template=template.template)
return body
def kube_create_job_object(name,
container_image,
namespace="default",
container_name="jobcontainer",
env_vars={}):
# Body is the object Body
body = client.V1Job(api_version="batch/v1", kind="Job")
# Body needs Metadata
# Attention: Each JOB must have a different name!
body.metadata = client.V1ObjectMeta(namespace=namespace, name=name)
# And a Status
body.status = client.V1JobStatus()
# Now we start with the Template...
template = client.V1PodTemplate()
template.template = client.V1PodTemplateSpec()
# Passing Arguments in Env:
env_list = []
for env_name, env_value in env_vars.items():
env_list.append(client.V1EnvVar(name=env_name, value=env_value))
container = client.V1Container(name=container_name,
image=container_image,
env=env_list)
template.template.spec = client.V1PodSpec(containers=[container],
restart_policy='Never')
# And finaly we can create our V1JobSpec!
body.spec = client.V1JobSpec(ttl_seconds_after_finished=30,
template=template.template)
return body
def _create_job_object(self,
name: str,
container_image: str,
namespace: str = None,
container_name: str = "servicecontainer",
env_vars: dict = {},
command: list = [],
active_deadline_seconds: int = 3600):
namespace = self._get_namespace(namespace)
body = client.V1Job(api_version="batch/v1", kind="Job")
body.metadata = client.V1ObjectMeta(namespace=namespace, name=name)
body.status = client.V1JobStatus()
template = client.V1PodTemplate()
template.template = client.V1PodTemplateSpec()
env_list = []
for env_name, env_value in env_vars.items():
env_list.append(client.V1EnvVar(name=env_name, value=env_value))
container = client.V1Container(name=container_name,
image=container_image,
env=env_list,
command=command)
template.template.spec = client.V1PodSpec(containers=[container],
restart_policy='Never')
# Set active_deadline_seconds
body.spec = client.V1JobSpec(
ttl_seconds_after_finished=600,
template=template.template,
active_deadline_seconds=active_deadline_seconds)
return body
def create_notification_job(zip_code: int) -> client.V1Job:
"""Creates job to fulfill requested notification"""
return bot.k8s_batch.create_namespaced_job(
namespace=namespace,
body=client.V1Job(
api_version='batch/v1',
kind='Job',
metadata=client.V1ObjectMeta(generate_name=JOB_NAME_PREFIX),
spec=client.V1JobSpec(
ttl_seconds_after_finished=JOB_TTL_SECONDS_AFTER_FINISHED,
backoff_limit=JOB_MAX_RETRIES,
template=client.V1PodTemplateSpec(spec=client.V1PodSpec(
restart_policy=JOB_RESTART_POLICY,
containers=[
client.V1Container(
name='worker',
image=job_image,
resources=client.V1ResourceRequirements(
requests=JOB_RESOURCE_REQUESTS),
args=['--worker', '--zip_code',
str(zip_code)],
env=[
client.V1EnvVar(name=key, value=value)
for key, value in {
MONGO_USER: mongodb_user,
MONGO_PASSWORD: mongodb_password,
MONGO_HOST: mongodb_host,
MONGO_PORT: mongodb_port,
MY_TURN_API_KEY: my_turn_api_key
}.items()
])
])))))
def create_job_object(job_name, instance_id):
# Configureate Pod template container
container = kube_cli.V1Container(
name="pi",
image="perl",
command=["perl", "-Mbignum=bpi", "-wle", "print bpi(2000)"])
# Create and configurate a spec section
template = kube_cli.V1PodTemplateSpec(
metadata=kube_cli.V1ObjectMeta(labels={"app": "pi"},
annotations={
'ai-flow/watched': 'True',
'ai-flow/job-id': str(instance_id),
'ai-flow/workflow-id': str(1)
}),
spec=kube_cli.V1PodSpec(restart_policy="Never",
containers=[container]))
# Create the specification of deployment
spec = kube_cli.V1JobSpec(template=template, backoff_limit=4)
# Instantiate the job object
job = kube_cli.V1Job(api_version="batch/v1",
kind="Job",
metadata=kube_cli.V1ObjectMeta(name=job_name),
spec=spec)
return job
def create_clingo_job(prog_id):
from kubernetes import config
config.load_incluster_config()
from kubernetes import client
api_instance = client.BatchV1Api()
namespace = 'default'
body = client.V1Job(
api_version='batch/v1',
kind='Job',
metadata=client.V1ObjectMeta(name='job-{}'.format(prog_id), ),
spec=client.V1JobSpec(template=client.V1PodTemplateSpec(
metadata=client.V1ObjectMeta(name='job-{}'.format(prog_id), ),
spec=client.V1PodSpec(
restart_policy='Never',
containers=[
client.V1Container(
name='aspaas-app',
image='gcr.io/{}/worker'.format(
current_app.config['PROJECT_ID']),
command=["python"],
args=["run.py", str(prog_id)],
)
],
),
), ),
)
api_instance.create_namespaced_job(namespace, body)
def create(self):
job_spec_name = "{}spec".format(self.name)
return client.V1JobSpec(
template=client.V1PodTemplateSpec(
metadata=client.V1ObjectMeta(name=job_spec_name),
spec=self.create_pod_spec()
)
)
def request_trader(self, stockid):
config.load_incluster_config()
v1 = client.BatchV1Api()
job = client.V1Job()
job.metadata = client.V1ObjectMeta(name="trader{}".format(stockid))
container = client.V1Container(name="trader{}".format(stockid),
image="outk/trader:v0.0.1",
image_pull_policy="IfNotPresent")
container.env = [
V1EnvVar(name="STOCK_ID", value=stockid),
V1EnvVar(name="TRADE_PWD", value=TRADE_PWD),
V1EnvVar(name="LOGIN_ID", value=LOGIN_ID),
V1EnvVar(name="LOGIN_PASS", value=LOGIN_PASS),
V1EnvVar(name="LOGIN_PAGE_ID", value=LOGIN_PAGE_ID),
V1EnvVar(name="LOGIN_CONTROL_ID", value=LOGIN_CONTROL_ID),
V1EnvVar(name="GET_PRICE_PAGE_ID", value=GET_PRICE_PAGE_ID),
V1EnvVar(name="GET_PRICE_DATA_STORE_ID",
value=GET_PRICE_DATA_STORE_ID),
V1EnvVar(name="GET_PRICE_CONTROL_ID", value=GET_PRICE_CONTROL_ID),
V1EnvVar(name="GET_BUYING_POWER_PAGE_ID",
value=GET_BUYING_POWER_PAGE_ID),
V1EnvVar(name="GET_BUYING_POWER_DATA_STORE_ID",
value=GET_BUYING_POWER_DATA_STORE_ID),
V1EnvVar(name="GET_BUYING_POWER_CONTROL_ID",
value=GET_BUYING_POWER_CONTROL_ID),
V1EnvVar(name="GET_STOCK_HOLDINGS_PAGE_ID",
value=GET_STOCK_HOLDINGS_PAGE_ID),
V1EnvVar(name="GET_STOCK_HOLDINGS_DATA_STORE_ID",
value=GET_STOCK_HOLDINGS_DATA_STORE_ID),
V1EnvVar(name="GET_STOCK_HOLDINGS_CONTROL_ID",
value=GET_STOCK_HOLDINGS_CONTROL_ID),
V1EnvVar(name="BUY_ORDER_PAGE_ID", value=BUY_ORDER_PAGE_ID),
V1EnvVar(name="BUY_ORDER_CONTROL_ID", value=BUY_ORDER_CONTROL_ID),
V1EnvVar(name="SELL_ORDER_PAGE_ID", value=SELL_ORDER_PAGE_ID),
V1EnvVar(name="SELL_ORDER_CONTROL_ID",
value=SELL_ORDER_CONTROL_ID),
V1EnvVar(name="ORDER_CANCEL_PAGE_ID", value=ORDER_CANCEL_PAGE_ID),
V1EnvVar(name="ORDER_CANCEL_CONTROL_ID",
value=ORDER_CANCEL_CONTROL_ID)
]
spec = client.V1PodSpec(containers=[container], restart_policy="Never")
podtemplatespec = client.V1PodTemplateSpec(
metadata=client.V1ObjectMeta(name="trader{}".format(stockid)),
spec=spec)
jobspec = client.V1JobSpec(template=podtemplatespec,
ttl_seconds_after_finished=12 * 60 * 60)
job.spec = jobspec
v1.create_namespaced_job(namespace="trader", body=job)
logging.info("trader{} job created.".format(stockid))
return
def create_generator_job(kubeconfigpath,serviceid,namespace,archeplaydatapath,generatorpath):
try:
config.load_kube_config("/home/app/web/kubeconfig")
batch_v1 = client.BatchV1Api()
volume1=client.V1Volume(
name="generatorjob"+serviceid,
host_path={"path": "/var/run"}
)
volume2=client.V1Volume(
name="kubeconfig",
host_path={"path": kubeconfigpath}
)
volume3=client.V1Volume(
name="archeplaydata",
host_path={"path":archeplaydatapath}
)
mount1 = client.V1VolumeMount(
name="generatorjob"+serviceid,
mount_path="/var/run"
)
mount2 = client.V1VolumeMount(
name="kubeconfig",
mount_path="/home/app/web/kubeconfig"
)
mount3 = client.V1VolumeMount(
name="archeplaydata",
mount_path="/home/app/web/archeplay/data"
)
container = client.V1Container(
name="generatorjob"+serviceid,
image="python:3.8.1-slim-buster",
volume_mounts=[mount1,mount2,mount3],
command=["bash",generatorpath],
image_pull_policy="Always"
)
# Create and configurate a spec section
template = client.V1PodTemplateSpec(
metadata=client.V1ObjectMeta(labels={"generatorjob": "generatorjob"+serviceid}),
spec=client.V1PodSpec(restart_policy="Never", containers=[container],volumes=[volume1,volume2,volume3]))
# Create the specification of deployment
spec = client.V1JobSpec(
template=template,
backoff_limit=0
)
# Instantiate the job object
job = client.V1Job(
api_version="batch/v1",
kind="Job",
metadata=client.V1ObjectMeta(name="generatorjob"+serviceid),
spec=spec)
api_response = batch_v1.create_namespaced_job(
body=job,
namespace=namespace)
success_message = "Generator Job Intitated"
return("success",success_message,str(api_response.status))
except Exception as Error:
error_message="Generator job Failed to Intitate Deploy Job"
return("error",error_message,str(Error))
def dispatch(endpoint, access_key, secret_key, bucket_name, object_name):
job_name = f"local-rebuild-{uuid.uuid1()}"
downloader_env = [
client.V1EnvVar(name="ENDPOINT", value=endpoint),
client.V1EnvVar(name="ACCESS_KEY", value=access_key),
client.V1EnvVar(name="SECRET_KEY", value=secret_key),
client.V1EnvVar(name="BUCKET_NAME", value=bucket_name),
client.V1EnvVar(name="OBJECT_NAME", value=object_name)
]
downloader_container = client.V1Container(
name="downloader",
image=os.getenv("DOWNLOADER_IMAGE"),
env=downloader_env,
volume_mounts=[client.V1VolumeMount(
name="processor-input", mount_path="/output")])
processor_container = client.V1Container(
name="processor",
image=os.getenv("PROCESSOR_IMAGE"),
volume_mounts=[client.V1VolumeMount(name="processor-input", mount_path="/input", read_only=True),
client.V1VolumeMount(name="processor-output", mount_path="/output")])
pod_spec = client.V1PodSpec(
restart_policy="Never",
init_containers=[downloader_container],
containers=[processor_container],
volumes=[
client.V1Volume(name="processor-input",
empty_dir=client.V1EmptyDirVolumeSource()),
client.V1Volume(name="processor-output",
empty_dir=client.V1EmptyDirVolumeSource()),
])
# Create and configure a spec section
template = client.V1PodTemplateSpec(
metadata=client.V1ObjectMeta(name=job_name, labels={
"app": "local-rebuild-processor"}),
spec=pod_spec)
# Create the specification of the job
spec = client.V1JobSpec(
template=template,
backoff_limit=0)
# Instantiate the job object
job = client.V1Job(
api_version="batch/v1",
kind="Job",
metadata=client.V1ObjectMeta(name=job_name, labels={
"app": "local-rebuild-processor"}),
spec=spec)
client.BatchV1Api().create_namespaced_job(
body=job,
namespace="default")
def create_job_object(container):
template = kubernetes_client.V1PodTemplateSpec(
metadata=kubernetes_client.V1ObjectMeta(labels={"app": "pi"}),
spec=kubernetes_client.V1PodSpec(restart_policy="Never",
containers=[container]))
return kubernetes_client.V1Job(
api_version="batch/v1",
kind="Job",
metadata=kubernetes_client.V1ObjectMeta(name="pi"),
spec=kubernetes_client.V1JobSpec(template=template, backoff_limit=4))
def calc(n):
if n < 2:
return 1
else:
config.load_incluster_config()
core_api = client.CoreV1Api()
batch_api = client.BatchV1Api()
watch_api = watch.Watch()
container = client.V1Container(name=name,
image=image,
args=["--calc", f"{n}"])
template = client.V1PodTemplateSpec(
spec=client.V1PodSpec(restart_policy="Never",
service_account_name=service_account,
containers=[container]))
job = client.V1Job(metadata=client.V1ObjectMeta(
namespace=namespace, generate_name=f"{name}-"),
spec=client.V1JobSpec(template=template,
ttl_seconds_after_finished=0))
res = batch_api.create_namespaced_job(
body=job,
namespace=namespace,
)
job_name = res.metadata.name
for event in watch_api.stream(func=core_api.list_namespaced_pod,
namespace=namespace,
label_selector=f"job-name={job_name}",
timeout_seconds=1200):
if event["object"].status.phase == "Succeeded":
pod_name = event["object"].metadata.name
watch_api.stop()
break
if pod_name:
log = core_api.read_namespaced_pod_log(
name=pod_name,
namespace=namespace,
)
val = int(log)
batch_api.delete_namespaced_job(job_name,
namespace,
propagation_policy="Background")
return val