def docker_job(self, image, config=None,
input=None, output=None, msdir=None,
shared_memory='1gb', build_label=None,
**kw):
"""
Add a task to a stimela recipe
image : stimela cab name, e.g. 'cab/simms'
name : This name will be part of the name of the contaier that will
execute the task (now optional)
config : Dictionary of options to parse to the task. This will modify
the parameters in the default parameter file which
can be viewd by running 'stimela cabs -i <cab name>', e.g 'stimela cabs -i simms'
input : input dirctory for cab
output : output directory for cab
msdir : MS directory for cab. Only specify if different from recipe ms_dir
"""
# check if name has any offending charecters
offenders = re.findall('\W', self.name)
if offenders:
raise StimelaCabParameterError('The cab name \'{:s}\' has some non-alphanumeric characters.'
' Charecters making up this name must be in [a-z,A-Z,0-9,_]'.format(self.name))
# Update I/O with values specified on command line
# TODO (sphe) I think this feature should be removed
script_context = self.recipe.stimela_context
input = script_context.get('_STIMELA_INPUT', None) or input
output = script_context.get('_STIMELA_OUTPUT', None) or output
output = os.path.abspath(output)
msdir = script_context.get('_STIMELA_MSDIR', None) or msdir
build_label = script_context.get(
'_STIMELA_BUILD_LABEL', None) or build_label
# Get location of template parameters file
cabs_logger = get_cabs(
'{0:s}/{1:s}_stimela_logfile.json'.format(stimela.LOG_HOME, build_label))
try:
cabpath = cabs_logger['{0:s}_{1:s}'.format(
build_label, image)]['DIR']
except KeyError:
raise StimelaCabParameterError(
'Cab {} has is uknown to stimela. Was it built?'.format(image))
parameter_file = cabpath+'/parameters.json'
name = '{0}-{1}{2}'.format(self.name, id(image),
str(time.time()).replace('.', ''))
_cab = cab.CabDefinition(indir=input, outdir=output,
msdir=msdir, parameter_file=parameter_file)
cont = docker.Container(image, name,
label=self.label, logger=self.log,
shared_memory=shared_memory,
log_container=stimela.LOG_FILE,
time_out=self.time_out)
# Container parameter file will be updated and validated before the container is executed
cont._cab = _cab
cont.parameter_file_name = '{0}/{1}.json'.format(
self.recipe.parameter_file_dir, name)
# Remove dismissable kw arguments:
ops_to_pop = []
for op in config:
if isinstance(config[op], dismissable):
ops_to_pop.append(op)
for op in ops_to_pop:
arg = config.pop(op)()
if arg is not None:
config[op] = arg
cont.config = config
cont.add_volume(
"{0:s}/cargo/cab/docker_run".format(self.recipe.stimela_path), "/docker_run", perm="ro")
cont.COMMAND = "/bin/sh -c /docker_run"
# These are standard volumes and
# environmental variables. These will be
# always exist in a cab container
cont.add_volume(self.recipe.stimela_path,
'/scratch/stimela', perm='ro')
cont.add_volume(self.recipe.parameter_file_dir, '/configs', perm='ro')
cont.add_environ('CONFIG', '/configs/{}.json'.format(name))
cab.IODEST = CONT_IO["docker"]
if msdir:
md = cab.IODEST["msfile"]
cont.add_volume(msdir, md)
cont.add_environ('MSDIR', md)
# Keep a record of the content of the
# volume
dirname, dirs, files = [a for a in next(os.walk(msdir))]
cont.msdir_content = {
"volume": dirname,
"dirs": dirs,
"files": files,
}
self.log.debug(
'Mounting volume \'{0}\' from local file system to \'{1}\' in the container'.format(msdir, md))
if input:
cont.add_volume(input, cab.IODEST["input"], perm='ro')
cont.add_environ('INPUT', cab.IODEST["input"])
# Keep a record of the content of the
# volume
dirname, dirs, files = [a for a in next(os.walk(input))]
cont.input_content = {
"volume": dirname,
"dirs": dirs,
"files": files,
}
self.log.debug('Mounting volume \'{0}\' from local file system to \'{1}\' in the container'.format(
input, cab.IODEST["input"]))
if not os.path.exists(output):
os.mkdir(output)
od = cab.IODEST["output"]
cont.add_environ('HOME', od)
cont.add_environ('OUTPUT', od)
self.log_dir = os.path.abspath(self.log_dir or output)
log_dir_name = os.path.basename(self.log_dir or output)
logfile_name = 'log-{0:s}.txt'.format(name.split('-')[0])
self.logfile = cont.logfile = '{0:s}/{1:s}'.format(
self.log_dir, logfile_name)
cont.add_volume(output, od, "rw")
if not os.path.exists(self.logfile):
with open(self.logfile, "w") as std:
pass
cont.add_volume(
self.logfile, "{0:s}/logfile".format(self.log_dir), "rw")
cont.add_environ('LOGFILE', "{0:}/logfile".format(self.log_dir))
self.log.debug(
'Mounting volume \'{0}\' from local file system to \'{1}\' in the container'.format(output, od))
cont.image = '{0}_{1}'.format(build_label, image)
# Added and ready for execution
self.job = cont
return 0
def singularity_job(self,
image,
config,
singularity_image_dir,
input=None,
output=None,
msdir=None,
**kw):
"""
Run task in singularity
image : stimela cab name, e.g. 'cab/simms'
name : This name will be part of the name of the contaier that will
execute the task (now optional)
config : Dictionary of options to parse to the task. This will modify
the parameters in the default parameter file which
can be viewd by running 'stimela cabs -i <cab name>', e.g 'stimela cabs -i simms'
input : input dirctory for cab
output : output directory for cab
msdir : MS directory for cab. Only specify if different from recipe ms_dir
"""
# check if name has any offending charecters
offenders = re.findall('\W', self.name)
if offenders:
raise StimelaCabParameterError(
'The cab name \'{:s}\' has some non-alphanumeric characters.'
' Charecters making up this name must be in [a-z,A-Z,0-9,_]'.
format(self.name))
# Update I/O with values specified on command line
# TODO (sphe) I think this feature should be removed
script_context = self.recipe.stimela_context
input = script_context.get('_STIMELA_INPUT', None) or input
output = script_context.get('_STIMELA_OUTPUT', None) or output
msdir = script_context.get('_STIMELA_MSDIR', None) or msdir
# Get location of template parameters file
cabpath = self.recipe.stimela_path + \
"/cargo/cab/{0:s}/".format(image.split("/")[1]) if not self.cabpath else \
os.path.join(self.cabpath, image.split("/")[1])
parameter_file = cabpath + '/parameters.json'
name = '{0}-{1}{2}'.format(self.name, id(image),
str(time.time()).replace('.', ''))
_cab = cab.CabDefinition(indir=input,
outdir=output,
msdir=msdir,
parameter_file=parameter_file)
cab.IODEST = CONT_IO["singularity"]
self.setup_job_log()
cont = singularity.Container(image,
name,
logger=self.log,
time_out=self.time_out)
# Container parameter file will be updated and validated before the container is executed
cont._cab = _cab
cont.parameter_file_name = '{0}/{1}.json'.format(
self.recipe.parameter_file_dir, name)
# Remove dismissable kw arguments:
ops_to_pop = []
for op in config:
if isinstance(config[op], dismissable):
ops_to_pop.append(op)
for op in ops_to_pop:
arg = config.pop(op)()
if arg is not None:
config[op] = arg
cont.config = config
# These are standard volumes and
# environmental variables. These will be
# always exist in a cab container
cont.add_volume(self.recipe.stimela_path,
'/scratch/stimela',
perm='ro')
cont.add_volume(cont.parameter_file_name,
'/scratch/configfile',
perm='ro',
noverify=True)
cont.add_volume(
"{0:s}/{1:s}/src/".format(
self.cabpath
or "{0:s}/cargo/cab".format(self.recipe.stimela_path),
_cab.task), "/scratch/code", "ro")
cont.add_volume(os.path.join(BIN, "stimela_singularity_run"),
"/singularity")
if msdir:
md = cab.IODEST["msfile"]
cont.add_volume(msdir, md)
# Keep a record of the content of the
# volume
dirname, dirs, files = [a for a in next(os.walk(msdir))]
cont.msdir_content = {
"volume": dirname,
"dirs": dirs,
"files": files,
}
self.log.debug(
'Mounting volume \'{0}\' from local file system to \'{1}\' in the container'
.format(msdir, md))
if input:
cont.add_volume(input, cab.IODEST["input"], perm='ro')
# Keep a record of the content of the
# volume
dirname, dirs, files = [a for a in next(os.walk(input))]
cont.input_content = {
"volume": dirname,
"dirs": dirs,
"files": files,
}
self.log.debug(
'Mounting volume \'{0}\' from local file system to \'{1}\' in the container'
.format(input, cab.IODEST["input"]))
if not os.path.exists(output):
os.mkdir(output)
od = cab.IODEST["output"]
self.log_dir = os.path.abspath(self.log_dir or output)
cont.logfile = self.tmp_logfile
cont.add_volume(cont.logfile, "/scratch/logfile", "rw")
cont.add_volume(output, od, "rw")
# temp files go into output
tmpfol = os.path.join(output, "tmp")
if not os.path.exists(tmpfol):
os.mkdir(tmpfol)
cont.add_volume(tmpfol, cab.IODEST["tmp"], "rw")
self.log.debug(
'Mounting volume \'{0}\' from local file system to \'{1}\' in the container'
.format(output, od))
simage = _cab.base.replace("/", "_")
cont.image = '{0:s}/{1:s}_{2:s}.img'.format(singularity_image_dir,
simage, _cab.tag)
# Added and ready for execution
self.job = cont
return 0
def udocker_job(self, image, config,
input=None, output=None, msdir=None,
**kw):
"""
Run task using udocker
image : stimela cab name, e.g. 'cab/simms'
name : This name will be part of the name of the contaier that will
execute the task (now optional)
config : Dictionary of options to parse to the task. This will modify
the parameters in the default parameter file which
can be viewd by running 'stimela cabs -i <cab name>', e.g 'stimela cabs -i simms'
input : input dirctory for cab
output : output directory for cab
msdir : MS directory for cab. Only specify if different from recipe ms_dir
"""
# check if name has any offending charecters
offenders = re.findall('\W', self.name)
if offenders:
raise StimelaCabParameterError('The cab name \'{:s}\' has some non-alphanumeric characters.'
' Charecters making up this name must be in [a-z,A-Z,0-9,_]'.format(self.name))
# Update I/O with values specified on command line
# TODO (sphe) I think this feature should be removed
script_context = self.recipe.stimela_context
input = script_context.get('_STIMELA_INPUT', None) or input
output = script_context.get('_STIMELA_OUTPUT', None) or output
msdir = script_context.get('_STIMELA_MSDIR', None) or msdir
# Get location of template parameters file
cabpath = self.recipe.stimela_path + \
"/cargo/cab/{0:s}/".format(image.split("/")[1])
parameter_file = cabpath+'/parameters.json'
name = '{0}-{1}{2}'.format(self.name, id(image),
str(time.time()).replace('.', ''))
_cab = cab.CabDefinition(indir=input, outdir=output,
msdir=msdir, parameter_file=parameter_file)
cab.IODEST = CONT_IO["udocker"]
cont = udocker.Container(image, name,
logger=self.log, time_out=self.time_out)
cont.add_volume(
"{0:s}/cargo/cab/docker_run".format(self.recipe.stimela_path), "/udocker_run")
cont.COMMAND = "/bin/sh -c /udocker_run"
# Container parameter file will be updated and validated before the container is executed
cont._cab = _cab
cont.parameter_file_name = '{0}/{1}.json'.format(
self.recipe.parameter_file_dir, name)
# Remove dismissable kw arguments:
ops_to_pop = []
for op in config:
if isinstance(config[op], dismissable):
ops_to_pop.append(op)
for op in ops_to_pop:
arg = config.pop(op)()
if arg is not None:
config[op] = arg
cont.config = config
# These are standard volumes and
# environmental variables. These will be
# always exist in a cab container
cont.add_volume(self.recipe.stimela_path, '/scratch/stimela')
cont.add_volume(cont.parameter_file_name,
'/scratch/configfile', noverify=True)
cont.add_volume("{0:s}/cargo/cab/{1:s}/src/".format(
self.recipe.stimela_path, _cab.task), "/scratch/code")
cont.add_environ('CONFIG', '/scratch/configfile'.format(name))
if msdir:
md = cab.IODEST["msfile"]
cont.add_volume(msdir, md)
cont.add_environ("MSDIR", md)
# Keep a record of the content of the
# volume
dirname, dirs, files = [a for a in next(os.walk(msdir))]
cont.msdir_content = {
"volume": dirname,
"dirs": dirs,
"files": files,
}
self.log.debug(
'Mounting volume \'{0}\' from local file system to \'{1}\' in the container'.format(msdir, md))
if input:
cont.add_volume(input, cab.IODEST["input"])
cont.add_environ("INPUT", cab.IODEST["input"])
# Keep a record of the content of the
# volume
dirname, dirs, files = [a for a in next(os.walk(input))]
cont.input_content = {
"volume": dirname,
"dirs": dirs,
"files": files,
}
self.log.debug('Mounting volume \'{0}\' from local file system to \'{1}\' in the container'.format(
input, cab.IODEST["input"]))
if not os.path.exists(output):
os.mkdir(output)
od = cab.IODEST["output"]
cont.WORKDIR = od
self.log_dir = os.path.abspath(self.log_dir or output)
log_dir_name = os.path.basename(self.log_dir or output)
logfile_name = 'log-{0:s}.txt'.format(name.split('-')[0])
self.logfile = cont.logfile = '{0:s}/{1:s}'.format(
self.log_dir, logfile_name)
if not os.path.exists(self.logfile):
with open(self.logfile, 'w') as std:
pass
cont.add_environ("LOGFILE", "/scratch/logfile")
cont.add_volume(self.logfile, "/scratch/logfile")
cont.add_volume(output, od)
cont.add_environ("OUTPUT", od)
self.log.debug(
'Mounting volume \'{0}\' from local file system to \'{1}\' in the container'.format(output, od))
if hasattr(cont._cab, "use_graphics") and cont._cab.use_graphics:
cont.use_graphics = True
cont.image = '{0:s}:{1:s}'.format(_cab.base, _cab.tag)
# Added and ready for execution
self.job = cont
return 0
def setup_job(self, image, config,
indir=None, outdir=None, msdir=None,
build_label=None, singularity_image_dir=None,
**kw):
"""
Setup job
image : stimela cab name, e.g. 'cab/simms'
name : This name will be part of the name of the contaier that will
execute the task (now optional)
config : Dictionary of options to parse to the task. This will modify
the parameters in the default parameter file which
can be viewd by running 'stimela cabs -i <cab name>', e.g 'stimela cabs -i simms'
indir : input dirctory for cab
outdir : output directory for cab
msdir : MS directory for cab. Only specify if different from recipe ms_dir
function : Python callable to execute
name : Name of function (if not given, will used function.__name__)
parameters : Parameters to parse to function
label : Function label; for logging purposes
"""
if self.jtype == "python":
self.image = image.__name__
if not callable(image):
raise utils.StimelaCabRuntimeError(
'Object given as function is not callable')
if self.name is None:
self.name = image.__name__
self.job = {
'function': image,
'parameters': config,
}
return 0
# check if name has any offending characters
offenders = re.findall('[^\w .-]', self.name)
if offenders:
raise StimelaCabParameterError('The cab name \'{:s}\' contains invalid characters.'
' Allowed charcaters are alphanumeric, plus [-_. ].'.format(self.name))
# Update I/O with values specified on command line
script_context = self.recipe.stimela_context
indir = script_context.get('_STIMELA_INPUT', None) or indir
outdir = script_context.get('_STIMELA_OUTPUT', None) or outdir
msdir = script_context.get('_STIMELA_MSDIR', None) or msdir
build_label = script_context.get(
'_STIMELA_BUILD_LABEL', None) or build_label
self.setup_job_log()
# make name palatable as container name
pausterized_name = re.sub("[\W]", "_", self.name)
name = '{0}-{1}{2}'.format(pausterized_name, id(image),
str(time.time()).replace('.', ''))
cont = getattr(CONT_MOD[self.jtype], "Container")(image, name,
logger=self.log,
workdir=WORKDIR,
time_out=self.time_out)
if self.jtype == "docker":
# Get location of template parameters file
cabs_loc = f"{stimela.LOG_HOME}/{build_label}_stimela_logfile.json"
cabs_logger = get_cabs(cabs_loc)
try:
cabpath = cabs_logger[f'{build_label}_{cont.image}']['DIR']
except KeyError:
main.build(["--us-only", cont.image.split("/")[-1],
"--no-cache", "--build-label", build_label])
cabs_logger = get_cabs(cabs_loc)
cabpath = cabs_logger[f'{build_label}_{cont.image}']['DIR']
else:
cabpath = os.path.join(CAB_PATH, image.split("/")[1])
# In case the user specified a custom cab
cabpath = os.path.join(self.cabpath, image.split("/")[1]) if self.cabpath else cabpath
parameter_file = os.path.join(cabpath, 'parameters.json')
_cab = cab.CabDefinition(indir=indir, outdir=outdir,
msdir=msdir, parameter_file=parameter_file)
cont.IODEST = CONT_IO
cont.cabname = _cab.task
if self.jtype == "docker":
cont.image = '{0}_{1}'.format(build_label, image)
elif self.jtype == "singularity":
simage = _cab.base.replace("/", "_")
if singularity_image_dir is None:
singularity_image_dir = os.path.join(".", "stimela_singularity_images")
cont.image = '{0:s}/{1:s}_{2:s}{3:s}'.format(singularity_image_dir,
simage, _cab.tag, singularity.suffix)
if not os.path.exists(cont.image):
main.pull(f"-s -cb {cont.cabname} -pf {singularity_image_dir}".split())
else:
cont.image = ":".join([_cab.base, _cab.tag])
# Container parameter file will be updated and validated before the container is executed
cont._cab = _cab
cont.parameter_file_name = '{0}/{1}.json'.format(
self.recipe.parameter_file_dir, name)
self.image = str(cont.image)
# Remove dismissable kw arguments:
ops_to_pop = []
for op in config:
if isinstance(config[op], dismissable):
ops_to_pop.append(op)
for op in ops_to_pop:
arg = config.pop(op)()
if arg is not None:
config[op] = arg
cont.config = config
# These are standard volumes and
# environmental variables. These will be
# always exist in a cab container
cont.add_volume(self.recipe.stimela_path,
'/scratch/stimela', perm='ro')
cont.add_volume(cont.parameter_file_name,
'/scratch/configfile', perm='ro', noverify=True)
cont.add_volume(os.path.join(cabpath, "src"), "/scratch/code", "ro")
if self.jtype == "singularity":
cont.RUNSCRIPT = f"/{self.jtype}"
else:
cont.RUNSCRIPT = f"/{self.jtype}_run"
cont.add_volume(f"{BIN}/stimela_runscript",
cont.RUNSCRIPT, perm="ro")
cont.add_environ('CONFIG', '/scratch/configfile')
cont.add_environ('HOME', WORKDIR)
if msdir:
md = cont.IODEST["msfile"]
cont.add_volume(msdir, md)
cont.add_environ("MSDIR", md)
# Keep a record of the content of the
# volume
dirname, dirs, files = [a for a in next(os.walk(msdir))]
cont.msdir_content = {
"volume": dirname,
"dirs": dirs,
"files": files,
}
self.log.debug(
'Mounting volume \'{0}\' from local file system to \'{1}\' in the container'.format(msdir, md))
if indir:
cont.add_volume(indir, cont.IODEST["input"], perm='ro')
cont.add_environ("INPUT", cont.IODEST["input"])
# Keep a record of the content of the
# volume
dirname, dirs, files = [a for a in next(os.walk(indir))]
cont.input_content = {
"volume": dirname,
"dirs": dirs,
"files": files,
}
self.log.debug('Mounting volume \'{0}\' from local file system to \'{1}\' in the container'.format(
indir, cont.IODEST["input"]))
if not os.path.exists(outdir):
os.mkdir(outdir)
od = cont.IODEST["output"]
cont.logfile = self.logfile
cont.add_volume(outdir, od, "rw")
cont.add_environ("OUTPUT", od)
# temp files go into output
tmpfol = os.path.join(outdir, "tmp")
if not os.path.exists(tmpfol):
os.mkdir(tmpfol)
cont.add_volume(tmpfol, cont.IODEST["tmp"], "rw")
cont.add_environ("TMPDIR", cont.IODEST["tmp"])
self.log.debug(
'Mounting volume \'{0}\' from local file system to \'{1}\' in the container'.format(outdir, od))
# Added and ready for execution
self.job = cont
return 0
