def status(self):
"""
Get job status from HPC.
Returns
-------
status : (str)
Job status. Possible status are 'PENDING','RUNNING','NOT IN QUEUE'.
"""
hpc = HPCInterface()
stdout, stderr = hpc.qstat(printout=False)
queue = stdout.splitlines()
job_lines = grep_list(self.name, queue)
if job_lines == []:
status = 'NOT IN QUEUE'
elif len(job_lines) > 1:
raise ValueError(
f'More than one job named "{self.name}" has been found in queue:\n{stdout}'
)
else:
job_line = job_lines[0].split()
status = job_line[4]
if status == 'PD':
status = 'PENDING'
if status == 'R':
status = 'RUNNING'
if status == 'CG':
status = 'COMPLETED'
return status
def cancel_job(self):
"""Cancel job on HPC"""
hpc = HPCInterface()
job_id = self.job_id()
hpc.cancel_jobs(job_id)
return
def run_job(self, write_input=True, sync=True):
"""
Run job on HPC. Input files are automatically written and sync to HPC is performed.
Parameters
----------
write_input : (bool), optional
Write input file stored in "inputs" dictionary. The default is True.
sync : (bool), optional
Sync files to HPC before running. The default is True
Returns
-------
stdout : (str)
Output.
stderr : (str)
Error.
"""
if write_input:
self.write_input()
hpc = HPCInterface()
if sync:
self.sync_to_hpc()
stdout, stderr = hpc.sbatch(
path=self.path_in_hpc,
job_script_filename=self.job_script_filename)
return stdout, stderr
def sync_to_hpc(self, stdouts=False, exclude=None, dry_run=False):
"""
Sync job data from local machine to HPC
Parameters
----------
stdouts : (bool), optional
Return output and error strings. The default is False.
exclude : (list), optional
List of files to exclude in rsync. The default is None.
dry_run : (bool), optional
Perform dry run in rsync with --dry-run. The default is False.
Returns
-------
stdout : (str)
Output.
stderr : (str)
Error.
"""
hpc = HPCInterface()
abs_path = op.abspath(self.path)
localdir = abs_path
stdout, stderr = hpc.rsync_to_hpc(localdir=localdir,
remotedir=self.path_in_hpc,
exclude=exclude,
dry_run=dry_run)
if stdouts:
return stdout, stderr
else:
return
def __init__(self,jobs=None,path=None,name=None,sort_by_name=True):
"""
Class to store sets of calculations
Parameters
----------
jobs : (list), optional
List of Job objects belonging to the dataset. The default is None.
path : (str), optional
Path of dataset directory. If None the work dir is used if jobs is None,
else the commonpath between the jobs is used. The default is None.
name : (str), optional
Name to assign to dataset. The default is None. If None the folder name is used.
sort_by_name : (bool), optional
Sort list of jobs based on Job names. The default is True.
"""
if jobs:
path = op.commonpath([j.path for j in jobs])
self.path = op.abspath(path)
else:
self.path = op.abspath(path) if path else os.getcwd()
self.name = name if name else op.basename(self.path)
self.sort_by_name = sort_by_name
self.jobs = jobs
if jobs:
self._group_jobs()
if sort_by_name:
self.jobs = sorted(self.jobs, key=operator.attrgetter('name'))
self._localdir = HPCInterface().localdir
self._workdir = HPCInterface().workdir
self._path_relative = self.path.replace(self._localdir,'')
self.path_in_hpc = self._workdir + self._path_relative
def __init__(self,
path=None,
inputs=None,
job_settings=None,
outputs=None,
job_script_filename=None,
name=None):
"""
Class to control and organize inputs and outputs of a generic job.
Parameters
----------
path : (str), optional
Path where job is stored. The default is None. If None the work dir is used.
inputs : (dict), optional
Dictionary with input data. The default is None.
job_settings : (dict), optional
Dictionary with job settings. The default is None. Documentation in ScriptHandler class in slurm.job_script module
outputs : (dict), optional
Dictionary with output data. The default is None.
job_script_filename : (str), optional
Filename of job script. The default is taken from the key 'filename' in the job_settings in the config file.
name : (str)
Name of the job. If none the name is searched in the job script.
"""
self.path = path if path else os.getcwd()
self.inputs = inputs
self.job_settings = job_settings
self.outputs = outputs
self.job_script_filename = job_script_filename if job_script_filename else ScriptHandler(
).filename
self._localdir = HPCInterface().localdir
self._workdir = HPCInterface().workdir
self._path_relative = op.abspath(self.path).replace(self._localdir, '')
self.path_in_hpc = self._workdir + self._path_relative
if outputs:
self.get_output_properties()
if name:
self.name = name
elif self.job_settings:
self.name = self.job_settings['name']
elif op.isfile(op.join(self.path, self.job_script_filename)):
s = ScriptHandler.from_file(self.path,
filename=self.job_script_filename)
self.name = s.settings['name']
else:
self.name = 'no_name'
if not self.job_settings:
self.job_settings = {}
self.job_settings['name'] = self.name
def job_queue(self):
"""
Print job queue from HPC on screen
Returns
-------
stdout : (str)
Output.
stderr : (str)
Error.
"""
hpc = HPCInterface()
stdout, stderr = hpc.qstat()
return stdout, stderr
def queue(self,stdouts=False):
"""
Display queue from HPC. If stdouts is True returns out and err strings.
Returns
-------
stdout : (str)
Output.
stderr : (str)
Error.
"""
hpc = HPCInterface()
stdout,stderr = hpc.qstat()
if stdouts:
return stdout,stderr
else:
return
def job_id(self):
"""Get job ID from the queue on HPC"""
hpc = HPCInterface()
stdout, stderr = hpc.qstat(printout=False)
queue = stdout.splitlines()
job_lines = grep_list(self.name, queue)
if job_lines == []:
raise ValueError(
f'Job named "{self.name}" is not currently running or pending')
elif len(job_lines) > 1:
raise ValueError(
f'More than one job named "{self.name}" has been found in queue:\n{stdout}'
)
else:
job_line = job_lines[0].split()
job_id = job_line[0]
return job_id