async def job_submission_list(
pagination: Pagination = Depends(),
all: Optional[bool] = Query(
None,
description=
"If supplied, do not limit job_submissions to only the current user",
),
slurm_job_ids: Optional[str] = Query(
None,
description=
"Comma-separated list of slurm-job-ids to match active job_submissions",
),
submit_status: Optional[JobSubmissionStatus] = Query(
None,
description="Limit results to those with matching status",
),
search: Optional[str] = Query(None),
sort_field: Optional[str] = Query(None),
sort_ascending: bool = Query(True),
token_payload: TokenPayload = Depends(
guard.lockdown(Permissions.JOB_SUBMISSIONS_VIEW)),
):
"""
List job_submissions for the authenticated user.
"""
logger.debug("Fetching job submissions")
identity_claims = IdentityClaims.from_token_payload(token_payload)
logger.debug(f"Extracted identity claims from token: {identity_claims}")
query = job_submissions_table.select()
logger.debug("Building query")
if submit_status:
query = query.where(job_submissions_table.c.status == submit_status)
if not all:
query = query.where(job_submissions_table.c.job_submission_owner_email
== identity_claims.user_email)
if slurm_job_ids is not None and slurm_job_ids != "":
try:
job_ids = [int(i) for i in slurm_job_ids.split(",")]
except Exception:
raise HTTPException(
status_code=status.HTTP_422_UNPROCESSABLE_ENTITY,
detail=
"Invalid slurm_job_ids param. Must be a comma-separated list of integers",
)
query = query.where(job_submissions_table.c.slurm_job_id.in_(job_ids))
if search is not None:
query = query.where(search_clause(search, searchable_fields))
if sort_field is not None:
query = query.order_by(
sort_clause(sort_field, sortable_fields, sort_ascending))
logger.debug(f"Query built as: {render_sql(query)}")
logger.debug("Awaiting query and response package")
response = await package_response(JobSubmissionResponse, query, pagination)
logger.debug(f"Response built as: {response}")
return response
async def job_submissions_agent_pending(token_payload: TokenPayload = Depends(
guard.lockdown(Permissions.JOB_SUBMISSIONS_VIEW)), ):
"""
Get a list of pending job submissions for the cluster-agent.
"""
identity_claims = IdentityClaims.from_token_payload(token_payload)
if identity_claims.cluster_id is None:
raise HTTPException(
status_code=status.HTTP_400_BAD_REQUEST,
detail=
"Access token does not contain a `cluster_id`. Cannot fetch pending submissions",
)
logger.info(
f"Fetching newly created job_submissions for cluster_id: {identity_claims.cluster_id}"
)
query = (select(columns=[
job_submissions_table.c.id,
job_submissions_table.c.job_submission_name,
job_submissions_table.c.job_submission_owner_email,
job_scripts_table.c.job_script_name,
job_scripts_table.c.job_script_data_as_string,
applications_table.c.application_name,
]).select_from(
job_submissions_table.join(job_scripts_table).join(applications_table)
).where(
job_submissions_table.c.status == JobSubmissionStatus.CREATED, ).where(
job_submissions_table.c.cluster_id == identity_claims.cluster_id, )
)
rows = await database.fetch_all(query)
return rows
async def applications_create(
application: ApplicationCreateRequest,
token_payload: TokenPayload = Depends(
guard.lockdown(Permissions.APPLICATIONS_EDIT)),
):
"""
Create new applications using an authenticated user token.
"""
identity_claims = IdentityClaims.from_token_payload(token_payload)
create_dict = dict(
**application.dict(exclude_unset=True),
application_owner_email=identity_claims.user_email,
)
try:
insert_query = applications_table.insert().returning(
applications_table)
application_data = await database.fetch_one(query=insert_query,
values=create_dict)
except INTEGRITY_CHECK_EXCEPTIONS as e:
raise HTTPException(status_code=status.HTTP_409_CONFLICT,
detail=str(e))
return application_data
async def applications_list(
user: bool = Query(False),
all: bool = Query(False),
search: Optional[str] = Query(None),
sort_field: Optional[str] = Query(None),
sort_ascending: bool = Query(True),
pagination: Pagination = Depends(),
token_payload: TokenPayload = Depends(
guard.lockdown(Permissions.APPLICATIONS_VIEW)),
):
"""
List all applications.
"""
identity_claims = IdentityClaims.from_token_payload(token_payload)
query = applications_table.select()
if user:
query = query.where(applications_table.c.application_owner_email ==
identity_claims.user_email)
if not all:
query = query.where(
not_(applications_table.c.application_identifier.is_(None)))
if search is not None:
query = query.where(search_clause(search, searchable_fields))
if sort_field is not None:
query = query.order_by(
sort_clause(sort_field, sortable_fields, sort_ascending))
return await package_response(ApplicationResponse, query, pagination)
async def job_script_list(
pagination: Pagination = Depends(),
all: Optional[bool] = Query(False),
search: Optional[str] = Query(None),
sort_field: Optional[str] = Query(None),
sort_ascending: bool = Query(True),
token_payload: TokenPayload = Depends(
guard.lockdown(Permissions.JOB_SCRIPTS_VIEW)),
):
"""
List job_scripts for the authenticated user.
Note::
Use responses instead of response_model to skip a second round of validation and serialization. This
is already happening in the ``package_response`` method. So, we uses ``responses`` so that FastAPI
can generate the correct OpenAPI spec but not post-process the response.
"""
query = job_scripts_table.select()
identity_claims = IdentityClaims.from_token_payload(token_payload)
if not all:
query = query.where(job_scripts_table.c.job_script_owner_email ==
identity_claims.user_email)
if search is not None:
query = query.where(search_clause(search, searchable_fields))
if sort_field is not None:
query = query.order_by(
sort_clause(sort_field, sortable_fields, sort_ascending))
return await package_response(JobScriptResponse, query, pagination)
async def job_submission_create(
job_submission: JobSubmissionCreateRequest,
token_payload: TokenPayload = Depends(
guard.lockdown(Permissions.JOB_SUBMISSIONS_EDIT)),
):
"""
Create a new job submission.
Make a post request to this endpoint with the required values to create a new job submission.
"""
identity_claims = IdentityClaims.from_token_payload(token_payload)
cluster_id = job_submission.cluster_id or identity_claims.cluster_id
if cluster_id is None:
raise HTTPException(
status_code=status.HTTP_400_BAD_REQUEST,
detail=
"Could not find a cluster_id in the request body or auth token.",
)
create_dict = dict(
**job_submission.dict(exclude_unset=True),
job_submission_owner_email=identity_claims.user_email,
status=JobSubmissionStatus.CREATED,
)
if job_submission.cluster_id is None:
create_dict.update(cluster_id=cluster_id)
exec_dir = create_dict.pop("execution_directory", None)
if exec_dir is not None:
create_dict.update(execution_directory=str(exec_dir))
select_query = job_scripts_table.select().where(
job_scripts_table.c.id == job_submission.job_script_id)
raw_job_script = await database.fetch_one(select_query)
if not raw_job_script:
raise HTTPException(
status_code=status.HTTP_404_NOT_FOUND,
detail=f"JobScript id={job_submission.job_script_id} not found.",
)
async with database.transaction():
try:
insert_query = job_submissions_table.insert().returning(
job_submissions_table)
job_submission_data = await database.fetch_one(query=insert_query,
values=create_dict)
except INTEGRITY_CHECK_EXCEPTIONS as e:
raise HTTPException(
status_code=status.HTTP_422_UNPROCESSABLE_ENTITY,
detail=str(e))
return job_submission_data
async def job_submission_agent_update(
job_submission_id: int,
new_status: str = Body(..., embed=True),
slurm_job_id: Optional[int] = Body(None, embed=True),
token_payload: TokenPayload = Depends(
guard.lockdown(Permissions.JOB_SUBMISSIONS_EDIT)),
):
"""
Update a job_submission with a new status.
Make a put request to this endpoint with the new status to update a job_submission.
"""
identity_claims = IdentityClaims.from_token_payload(token_payload)
if identity_claims.cluster_id is None:
logger.error("Access token does not contain a cluster_id")
raise HTTPException(
status_code=status.HTTP_400_BAD_REQUEST,
detail=
"Access token does not contain a `cluster_id`. Cannot update job_submission",
)
logger.info(f"Setting status to: {new_status} "
f"for job_submission: {job_submission_id} "
f"on cluster_id: {identity_claims.cluster_id}")
update_values: Dict[str, Any] = dict(status=new_status)
if slurm_job_id is not None:
update_values.update(slurm_job_id=slurm_job_id)
update_query = (job_submissions_table.update().where(
job_submissions_table.c.id == job_submission_id).where(
job_submissions_table.c.cluster_id == identity_claims.cluster_id).
values(**update_values).returning(job_submissions_table))
result = await database.fetch_one(update_query)
if result is None:
raise HTTPException(
status_code=status.HTTP_404_NOT_FOUND,
detail=(f"JobSubmission with id={job_submission_id} "
"and cluster_id={identity_claims.cluster_id} not found."),
)
return result
async def job_submissions_agent_active(token_payload: TokenPayload = Depends(
guard.lockdown(Permissions.JOB_SUBMISSIONS_VIEW)), ):
"""
Get a list of active job submissions for the cluster-agent.
"""
identity_claims = IdentityClaims.from_token_payload(token_payload)
if identity_claims.cluster_id is None:
logger.error("Access token does not contain a cluster_id")
raise HTTPException(
status_code=status.HTTP_400_BAD_REQUEST,
detail=
"Access token does not contain a `cluster_id`. Cannot fetch pending submissions",
)
logger.info(
f"Fetching active job_submissions for cluster_id: {identity_claims.cluster_id}"
)
query = (job_submissions_table.select().where(
job_submissions_table.c.status == JobSubmissionStatus.SUBMITTED).where(
job_submissions_table.c.cluster_id == identity_claims.cluster_id))
rows = await database.fetch_all(query)
return rows
job_submission_data = await database.fetch_one(query=insert_query,
values=create_dict)
except INTEGRITY_CHECK_EXCEPTIONS as e:
raise HTTPException(
status_code=status.HTTP_422_UNPROCESSABLE_ENTITY,
detail=str(e))
return job_submission_data
@router.get(
"/job-submissions/{job_submission_id}",
description="Endpoint to get a job_submission",
response_model=JobSubmissionResponse,
dependencies=[Depends(guard.lockdown(Permissions.JOB_SUBMISSIONS_VIEW))],
)
async def job_submission_get(job_submission_id: int = Query(...)):
"""
Return the job_submission given it's id.
"""
query = job_submissions_table.select().where(
job_submissions_table.c.id == job_submission_id)
job_submission_data = await database.fetch_one(query)
if not job_submission_data:
raise HTTPException(
status_code=404,
detail=f"JobSubmission with id={job_submission_id} not found.")
return job_submission_data
values=create_dict)
except INTEGRITY_CHECK_EXCEPTIONS as e:
raise HTTPException(status_code=status.HTTP_409_CONFLICT,
detail=str(e))
return application_data
@router.post(
"/applications/{application_id}/upload",
status_code=status.HTTP_201_CREATED,
description=(
"Endpoint for uploading application files. "
"The file should be a gzipped tar-file (e.g. `jobbergate.tar.gz`)."),
dependencies=[Depends(guard.lockdown(Permissions.APPLICATIONS_EDIT))],
)
async def applications_upload(
application_id: int = Query(
...,
description="id of the application for which to upload a file"),
upload_file: UploadFile = File(
...,
description="The gzipped application tar-file to be uploaded"),
content_length: int = Header(...),
):
"""
Upload application tarball using an authenticated user token.
"""
if content_length > settings.MAX_UPLOAD_FILE_SIZE:
raise HTTPException(
async def job_script_create(
job_script: JobScriptCreateRequest,
token_payload: TokenPayload = Depends(
guard.lockdown(Permissions.JOB_SCRIPTS_EDIT)),
):
"""
Create a new job script.
Make a post request to this endpoint with the required values to create a new job script.
"""
logger.debug(f"Creating job_script with: {job_script}")
select_query = applications_table.select().where(
applications_table.c.id == job_script.application_id)
raw_application = await database.fetch_one(select_query)
if not raw_application:
raise HTTPException(
status_code=status.HTTP_404_NOT_FOUND,
detail=
f"Application with id={job_script.application_id} not found.",
)
application = ApplicationResponse.parse_obj(raw_application)
logger.debug("Fetching application tarfile")
s3_application_tar = get_s3_object_as_tarfile(application.id)
identity_claims = IdentityClaims.from_token_payload(token_payload)
create_dict = dict(
**{
k: v
for (k, v) in job_script.dict(exclude_unset=True).items()
if k != "param_dict"
},
job_script_owner_email=identity_claims.user_email,
)
# Use application_config from the application as a baseline of defaults
print("APP CONFIG: ", application.application_config)
param_dict = safe_load(application.application_config)
# User supplied param dict is optional and may override defaults
param_dict.update(**job_script.param_dict)
logger.debug("Rendering job_script data as string")
job_script_data_as_string = build_job_script_data_as_string(
s3_application_tar, param_dict)
sbatch_params = create_dict.pop("sbatch_params", [])
create_dict["job_script_data_as_string"] = inject_sbatch_params(
job_script_data_as_string, sbatch_params)
logger.debug("Inserting job_script")
try:
insert_query = job_scripts_table.insert().returning(job_scripts_table)
job_script_data = await database.fetch_one(query=insert_query,
values=create_dict)
except INTEGRITY_CHECK_EXCEPTIONS as e:
raise HTTPException(status_code=status.HTTP_422_UNPROCESSABLE_ENTITY,
detail=str(e))
logger.debug(f"Created job_script={job_script_data}")
return job_script_data
job_script_data = await database.fetch_one(query=insert_query,
values=create_dict)
except INTEGRITY_CHECK_EXCEPTIONS as e:
raise HTTPException(status_code=status.HTTP_422_UNPROCESSABLE_ENTITY,
detail=str(e))
logger.debug(f"Created job_script={job_script_data}")
return job_script_data
@router.get(
"/job-scripts/{job_script_id}",
description="Endpoint to get a job_script",
response_model=JobScriptResponse,
dependencies=[Depends(guard.lockdown(Permissions.JOB_SCRIPTS_VIEW))],
)
async def job_script_get(job_script_id: int = Query(...)):
"""
Return the job_script given its id.
"""
query = job_scripts_table.select().where(
job_scripts_table.c.id == job_script_id)
job_script = await database.fetch_one(query)
if not job_script:
raise HTTPException(
status_code=status.HTTP_404_NOT_FOUND,
detail=f"JobScript with id={job_script_id} not found.",
)
return job_script
