def watch_jobs(self, job_db, app=None):
"""Use SSH connection to slurm submitnode to monitor jobs.
:param job_db: Dictionary which contains all running jobs.
"""
slurm_connection = SSHClient(
hostname=SlurmJobManagerCERN.SLURM_HEADNODE_HOSTNAME,
port=SlurmJobManagerCERN.SLURM_HEADNODE_PORT,
)
statuses_to_skip = ['succeeded', 'failed']
while True:
logging.debug('Starting a new stream request to watch Jobs')
try:
slurm_jobs = {}
for id, job_dict in job_db.items():
if (not job_db[id]['deleted']
and job_db[id]['compute_backend'] == 'slurmcern'
and not job_db[id]['status'] in statuses_to_skip):
slurm_jobs[job_dict['backend_job_id']] = id
if not slurm_jobs.keys():
logging.error('No slurm jobs')
continue
slurm_query_cmd = self.format_slurm_job_query(
slurm_jobs.keys())
stdout = slurm_connection.exec_command(slurm_query_cmd)
for item in stdout.rstrip().split('\n'):
slurm_job_status = item.split('|')[0]
slurm_job_id = item.split('|')[1]
job_id = slurm_jobs[slurm_job_id]
if slurm_job_status in slurmJobStatus['succeeded']:
SlurmJobManagerCERN.get_outputs()
job_db[job_id]['status'] = 'succeeded'
job_db[job_id]['deleted'] = True
job_db[job_id]['log'] = \
SlurmJobManagerCERN.get_logs(
backend_job_id=job_dict['backend_job_id'],
workspace=job_db[
job_id]['obj'].workflow_workspace)
store_logs(logs=job_db[job_id]['log'], job_id=job_id)
if slurm_job_status in slurmJobStatus['failed']:
SlurmJobManagerCERN.get_outputs()
job_db[job_id]['status'] = 'failed'
job_db[job_id]['deleted'] = True
job_db[job_id]['log'] = \
SlurmJobManagerCERN.get_logs(
backend_job_id=job_dict['backend_job_id'],
workspace=job_db[
job_id]['obj'].workflow_workspace)
store_logs(logs=job_db[job_id]['log'], job_id=job_id)
except Exception as e:
logging.error("Unexpected error: {}".format(e), exc_info=True)
time.sleep(120)
def watch_jobs(self, job_db, app=None):
"""Use SSH connection to slurm submitnode to monitor jobs.
:param job_db: Dictionary which contains all running jobs.
"""
slurm_connection = SSHClient(
hostname=self.job_manager_cls.SLURM_HEADNODE_HOSTNAME,
port=self.job_manager_cls.SLURM_HEADNODE_PORT,
)
statuses_to_skip = ["succeeded", "failed"]
while True:
logging.debug("Starting a new stream request to watch Jobs")
try:
slurm_jobs = {}
for id, job_dict in job_db.items():
if (not job_db[id]["deleted"]
and job_db[id]["compute_backend"] == "slurmcern"
and not job_db[id]["status"] in statuses_to_skip):
slurm_jobs[job_dict["backend_job_id"]] = id
if not slurm_jobs.keys():
logging.error("No slurm jobs")
continue
slurm_query_cmd = self.format_slurm_job_query(
slurm_jobs.keys())
stdout = slurm_connection.exec_command(slurm_query_cmd)
for item in stdout.rstrip().split("\n"):
slurm_job_status = item.split("|")[0]
slurm_job_id = item.split("|")[1]
job_id = slurm_jobs[slurm_job_id]
if slurm_job_status in slurmJobStatus["succeeded"]:
self.job_manager_cls.get_outputs()
job_db[job_id]["status"] = "succeeded"
job_db[job_id]["deleted"] = True
job_db[job_id]["log"] = self.job_manager_cls.get_logs(
backend_job_id=job_dict["backend_job_id"],
workspace=job_db[job_id]["obj"].workflow_workspace,
)
store_logs(logs=job_db[job_id]["log"], job_id=job_id)
if slurm_job_status in slurmJobStatus["failed"]:
self.job_manager_cls.get_outputs()
job_db[job_id]["status"] = "failed"
job_db[job_id]["deleted"] = True
job_db[job_id]["log"] = self.job_manager_cls.get_logs(
backend_job_id=job_dict["backend_job_id"],
workspace=job_db[job_id]["obj"].workflow_workspace,
)
store_logs(logs=job_db[job_id]["log"], job_id=job_id)
except Exception as e:
logging.error("Unexpected error: {}".format(e), exc_info=True)
time.sleep(120)
def execute(self):
"""Execute / submit a job with Slurm."""
self.cmd = self._encode_cmd(" ".join(self.cmd))
initialize_krb5_token(workflow_uuid=self.workflow_uuid)
self.slurm_connection = SSHClient(
hostname=SlurmJobManagerCERN.SLURM_HEADNODE_HOSTNAME,
port=SlurmJobManagerCERN.SLURM_HEADNODE_PORT,
)
self._transfer_inputs()
self._dump_job_file()
self._dump_job_submission_file()
stdout = self.slurm_connection.exec_command(
"cd {} && sbatch --parsable {}".format(
SlurmJobManagerCERN.SLURM_WORKSAPCE_PATH,
self.job_description_file))
backend_job_id = stdout.rstrip()
return backend_job_id
def get_outputs():
"""Transfer job outputs to REANA."""
os.chdir(SlurmJobManagerCERN.REANA_WORKSPACE_PATH)
slurm_connection = SSHClient()
sftp = slurm_connection.ssh_client.open_sftp()
SlurmJobManagerCERN._download_dir(
sftp, SlurmJobManagerCERN.SLURM_WORKSAPCE_PATH,
SlurmJobManagerCERN.REANA_WORKSPACE_PATH)
sftp.close()
class SlurmJobManagerCERN(JobManager):
"""Slurm job management."""
SLURM_WORKSAPCE_PATH = ""
"""Absolute path inside slurm head node used for submission."""
REANA_WORKSPACE_PATH = ""
"""Absolute REANA workspace path."""
SLURM_HEADNODE_HOSTNAME = os.getenv("SLURM_HOSTNAME", "hpc-batch.cern.ch")
"""Hostname of SLURM head-node used for job management via SSH."""
SLURM_HEADNODE_PORT = os.getenv("SLURM_CLUSTER_PORT", "22")
"""Port of SLURM head node."""
SLURM_PARTITION = os.getenv("SLURM_PARTITION", "batch-short")
"""Default slurm partition."""
SLURM_HOME_PATH = os.getenv("SLURM_HOME_PATH", "")
"""Default SLURM home path."""
def __init__(
self,
docker_img=None,
cmd=None,
prettified_cmd=None,
env_vars=None,
workflow_uuid=None,
workflow_workspace=None,
cvmfs_mounts="false",
shared_file_system=False,
job_name=None,
kerberos=False,
kubernetes_uid=None,
unpacked_img=False,
):
"""Instanciate Slurm job manager.
:param docker_img: Docker image.
:type docker_img: str
:param cmd: Command to execute.
:type cmd: list
:param prettified_cmd: pretified version of command to execute.
:type prettified_cmd: str
:param env_vars: Environment variables.
:type env_vars: dict
:param workflow_id: Unique workflow id.
:type workflow_id: str
:param workflow_workspace: Workflow workspace path.
:type workflow_workspace: str
:param cvmfs_mounts: list of CVMFS mounts as a string.
:type cvmfs_mounts: str
:param shared_file_system: if shared file system is available.
:type shared_file_system: bool
:param job_name: Name of the job
:type job_name: str
"""
super(SlurmJobManagerCERN, self).__init__(
docker_img=docker_img,
cmd=cmd,
prettified_cmd=prettified_cmd,
env_vars=env_vars,
workflow_uuid=workflow_uuid,
job_name=job_name,
)
self.compute_backend = "Slurm"
self.workflow_workspace = workflow_workspace
self.cvmfs_mounts = cvmfs_mounts
self.shared_file_system = shared_file_system
self.job_file = "job.sh"
self.job_description_file = "job_description.sh"
def _transfer_inputs(self):
"""Transfer inputs to SLURM submit node."""
stdout = self.slurm_connection.exec_command("pwd")
self.slurm_home_path = SlurmJobManagerCERN.SLURM_HOME_PATH or stdout.rstrip(
)
SlurmJobManagerCERN.SLURM_WORKSAPCE_PATH = os.path.join(
self.slurm_home_path, self.workflow_workspace[1:])
SlurmJobManagerCERN.REANA_WORKSPACE_PATH = self.workflow_workspace
self.slurm_connection.exec_command("mkdir -p {}".format(
SlurmJobManagerCERN.SLURM_WORKSAPCE_PATH))
sftp = self.slurm_connection.ssh_client.open_sftp()
os.chdir(self.workflow_workspace)
for dirpath, dirnames, filenames in os.walk(self.workflow_workspace):
try:
sftp.mkdir(os.path.join(self.slurm_home_path, dirpath[1:]))
except Exception:
pass
for file in filenames:
sftp.put(
os.path.join(dirpath, file),
os.path.join(self.slurm_home_path, dirpath[1:], file),
)
@JobManager.execution_hook
def execute(self):
"""Execute / submit a job with Slurm."""
self.cmd = self._encode_cmd(" ".join(self.cmd))
initialize_krb5_token(workflow_uuid=self.workflow_uuid)
self.slurm_connection = SSHClient(
hostname=SlurmJobManagerCERN.SLURM_HEADNODE_HOSTNAME,
port=SlurmJobManagerCERN.SLURM_HEADNODE_PORT,
)
self._transfer_inputs()
self._dump_job_file()
self._dump_job_submission_file()
stdout = self.slurm_connection.exec_command(
"cd {} && sbatch --parsable {}".format(
SlurmJobManagerCERN.SLURM_WORKSAPCE_PATH,
self.job_description_file))
backend_job_id = stdout.rstrip()
return backend_job_id
def _dump_job_submission_file(self):
"""Dump job submission file to the Slurm submit node."""
job_template = ("#!/bin/bash \n"
"#SBATCH --job-name={job_name} \n"
"#SBATCH --output=reana_job.%j.out \n"
"#SBATCH --error=reana_job.%j.err \n"
"#SBATCH --partition {partition} \n"
"#SBATCH --time 60 \n"
"srun {command}").format(
partition=SlurmJobManagerCERN.SLURM_PARTITION,
job_name=self.job_name,
command=self._wrap_singularity_cmd(),
)
self.slurm_connection.exec_command(
'cd {} && job="{}" && echo "$job"> {}'.format(
SlurmJobManagerCERN.SLURM_WORKSAPCE_PATH,
job_template,
self.job_description_file,
))
def _dump_job_file(self):
"""Dump job file."""
job_template = "#!/bin/bash \n{}".format(self.cmd)
self.slurm_connection.exec_command(
'cd {} && job="{}" && echo "$job" > {} && chmod +x {}'.format(
SlurmJobManagerCERN.SLURM_WORKSAPCE_PATH,
job_template,
self.job_file,
self.job_file,
))
def _encode_cmd(self, cmd):
"""Encode base64 cmd."""
encoded_cmd = base64.b64encode(cmd.encode("utf-8")).decode("utf-8")
return "echo {}|base64 -d|bash".format(encoded_cmd)
def _wrap_singularity_cmd(self):
"""Wrap command in singulrity."""
base_singularity_cmd = (
"singularity exec -B {SLURM_WORKSAPCE}:{REANA_WORKSPACE}"
" docker://{DOCKER_IMG} {CMD}".format(
SLURM_WORKSAPCE=SlurmJobManagerCERN.SLURM_WORKSAPCE_PATH,
REANA_WORKSPACE=SlurmJobManagerCERN.REANA_WORKSPACE_PATH,
DOCKER_IMG=self.docker_img,
CMD="./" + self.job_file,
))
return base_singularity_cmd
def get_outputs():
"""Transfer job outputs to REANA."""
os.chdir(SlurmJobManagerCERN.REANA_WORKSPACE_PATH)
slurm_connection = SSHClient()
sftp = slurm_connection.ssh_client.open_sftp()
SlurmJobManagerCERN._download_dir(
sftp,
SlurmJobManagerCERN.SLURM_WORKSAPCE_PATH,
SlurmJobManagerCERN.REANA_WORKSPACE_PATH,
)
sftp.close()
def _download_dir(sftp, remote_dir, local_dir):
"""Download remote directory content."""
os.path.exists(local_dir) or os.makedirs(local_dir)
dir_items = sftp.listdir_attr(remote_dir)
for item in dir_items:
remote_path = os.path.join(remote_dir, item.filename)
local_path = os.path.join(local_dir, item.filename)
if S_ISDIR(item.st_mode):
SlurmJobManagerCERN._download_dir(sftp, remote_path,
local_path)
else:
sftp.get(remote_path, local_path)
def get_logs(backend_job_id, workspace):
"""Return job logs if log files are present."""
stderr_file = os.path.join(workspace,
"reana_job." + str(backend_job_id) + ".err")
stdout_file = os.path.join(workspace,
"reana_job." + str(backend_job_id) + ".out")
log_files = [stderr_file, stdout_file]
job_log = ""
try:
for file in log_files:
with open(file, "r") as log_file:
job_log += log_file.read()
return job_log
except Exception as e:
msg = "Job logs of {} were not found. {}".format(backend_job_id, e)
logging.error(msg, exc_info=True)
return msg
