def main(ns):
"""
A generic plugin that uses Spark to:
read data,
transform data with given code,
and maybe write transformed data
Assume code is written in Python. For Scala or R code, use another option.
"""
job_id = ns.job_id
module = get_pymodule(ns.app_name)
pjob_id = api.parse_job_id(ns.app_name, job_id)
log_details = dict(
module_name=module.__name__,
app_name=ns.app_name, job_id=job_id)
conf, osenv, files, pyFiles = pyspark_context.get_spark_conf(ns.app_name)
conf['spark.app.name'] = "%s__%s" % (conf['spark.app.name'], job_id)
conf.update(ns.spark_conf)
osenv.update(ns.spark_env)
sc = pyspark_context.get_spark_context(
conf=conf, osenv=osenv, files=files, pyFiles=pyFiles)
apply_data_transform(
ns=ns, sc=sc, log_details=log_details, pjob_id=pjob_id, module=module)
sc.stop()
def main(ns):
"""
A generic plugin that schedules arbitrary bash jobs using Stolos
Assume code is written in Python. For Scala or R code, use another option.
"""
job_id = ns.job_id
ld = dict(app_name=ns.app_name, job_id=ns.job_id)
log.info('Running bash job', extra=ld)
cmd = get_bash_cmd(ns.app_name)
if ns.bash_cmd:
cmd += ' '.join(ns.bash_cmd)
log.debug("Appending user-supplied bash options to defaults",
extra=dict(app_name=ns.app_name, job_id=job_id, cmd=cmd))
ld.update(cmd=cmd)
if not cmd:
raise UserWarning(
"You need to specify bash options or configure default bash"
" options")
_cmdargs = dict(**ns.__dict__)
_cmdargs.update(api.parse_job_id(ns.app_name, job_id))
cmd = cmd.format(**_cmdargs)
if ns.redirect_to_stderr:
_std = sys.stderr
else:
_std = PIPE
log.info('running command', extra=ld)
returncode, stdout, stderr = run(cmd,
shell=True,
timeout=ns.watch,
stdout=_std,
stderr=_std)
ld = dict(bash_returncode=returncode, stdout=stdout, stderr=stderr, **ld)
if returncode == -9:
log_and_raise("Bash job timed out", ld)
elif returncode != 0:
# this raises an error and logs output:
log_and_raise("Bash job failed", ld)
else:
log.info("Bash job succeeded", extra=ld)
def main(ns):
"""
A generic plugin that schedules arbitrary bash jobs using Stolos
Assume code is written in Python. For Scala or R code, use another option.
"""
job_id = ns.job_id
ld = dict(app_name=ns.app_name, job_id=ns.job_id)
log.info('Running bash job', extra=ld)
cmd = get_bash_cmd(ns.app_name)
if ns.bash_cmd:
cmd += ' '.join(ns.bash_cmd)
log.debug(
"Appending user-supplied bash options to defaults", extra=dict(
app_name=ns.app_name, job_id=job_id, cmd=cmd))
ld.update(cmd=cmd)
if not cmd:
raise UserWarning(
"You need to specify bash options or configure default bash"
" options")
_cmdargs = dict(**ns.__dict__)
_cmdargs.update(api.parse_job_id(ns.app_name, job_id))
cmd = cmd.format(**_cmdargs)
if ns.redirect_to_stderr:
_std = sys.stderr
else:
_std = PIPE
log.info('running command', extra=ld)
returncode, stdout, stderr = run(
cmd, shell=True, timeout=ns.watch, stdout=_std, stderr=_std)
ld = dict(bash_returncode=returncode, stdout=stdout, stderr=stderr, **ld)
if returncode == -9:
log_and_raise("Bash job timed out", ld)
elif returncode != 0:
# this raises an error and logs output:
log_and_raise("Bash job failed", ld)
else:
log.info("Bash job succeeded", extra=ld)