def run(batch, runs, cwd=None):
"""Submit a job array for the given runs
"""
txt_output = text_file_directory(batch)
scripts = script_file_directory(batch)
if not os.path.exists(txt_output):
os.mkdir(txt_output)
os.chmod(txt_output, 0777)
if not os.path.exists(scripts):
os.mkdir(scripts)
os.chmod(scripts, 0755)
with bpcursor() as cursor:
bats = []
batch_array = BPBatchArray.create(cursor, batch)
for idx, run in enumerate(runs):
task_id = idx + TASK_ID_START
bats.append(
BPBatchArrayTask.create(cursor, batch_array, run, task_id))
script = "#!/bin/sh\n"
#
# A work-around if HOME has been defined differently on the host
#
script += """
if [ ! -z "$SGE_O_HOME" ]; then
export HOME="$SGE_O_HOME"
echo "Set home to $HOME"
fi
"""
#
# Source cpenv.sh
#
script += ". %s\n" % os.path.join(PREFIX, "bin", "cpenv.sh")
if JS_USE_REST:
#
# This is a REST PUT to JobStatus.py to create the job record
#
data = "{" + (",".join([
'"%s":"%s"' % (k, v)
for k, v in ((K_ACTION, A_UPDATE), (K_STATUS, JS_RUNNING))
]))
data += ',"%s":\'$JOB_ID\'' % JOB_ID
data += ',"%s":%d' % (BATCH_ARRAY_ID, batch_array.batch_array_id)
data += ',"%s":\'$SGE_TASK_ID\'' % TASK_ID
data += ',"%s":"\'$HOSTNAME\'"}' % (K_HOST_NAME)
script += """BATCHPROFILER_COMMAND=`curl -s """
script += """-H "Content-type: application/json" -X PUT """
script += """--data '%s' %s/JobStatus.py`\n""" % (
data, BATCHPROFILER_DEFAULTS[URL])
else:
script += """
if [ -e $HOME/.batchprofiler.sh ]; then
. $HOME/.batchprofiler.sh
fi
"""
script += "BATCHPROFILER_COMMAND="
script += "`PYTHONPATH=%s:$PYTHONPATH " % os.path.dirname(__file__)
script += "python -c 'from JobStatus import update_status;"
script += "print update_status(%d, '$JOB_ID', '$SGE_TASK_ID', " % \
batch_array.batch_array_id
script += "\"%s\", \"'$HOSTNAME'\")'`\n" % JS_RUNNING
#
# CD to the CellProfiler root directory
#
if cwd is None:
cwd = batch.cpcluster
script += 'cd %s\n' % cwd
#
# set +e allows the command to error-out without ending this script.
# This lets us capture the error status.
#
script += "set +e\n"
#
# Run CellProfiler
#
script += "echo 'Running '$BATCHPROFILER_COMMAND' on '$HOSTNAME\n"
script += "echo $BATCHPROFILER_COMMAND| bash\n"
#
# Figure out the status from the error code
#
script += "if [ $? == 0 ]; then\n"
script += "JOB_STATUS=%s\n" % JS_DONE
script += "else\n JOB_STATUS=%s\n " % JS_ERROR
script += "fi\n"
#
# Go back to erroring-out
#
script += "set -e\n"
if JS_USE_REST:
#
# Set the status based on the result from CellProfiler
# Use CURL again
#
script += """curl -s -H "Content-type: application/json" -X PUT """
script += """--data '{"action":"update","%s":'$JOB_ID',""" % JOB_ID
script += """"%s":%d,""" % (BATCH_ARRAY_ID, batch_array.batch_array_id)
script += """"%s":'$SGE_TASK_ID',""" % TASK_ID
script += """"%s":"'$JOB_STATUS'"}' """ % K_STATUS
script += '%s/JobStatus.py\n' % BATCHPROFILER_DEFAULTS[URL]
else:
script += "PYTHONPATH=%s:$PYTHONPATH " % os.path.dirname(__file__)
script += "python -c 'from JobStatus import update_status;"
script += "print update_status(%d, '$JOB_ID', '$SGE_TASK_ID', " % \
batch_array.batch_array_id
script += "\"'$JOB_STATUS'\", \"'$HOSTNAME'\")'\n"
script_filename = batch_array_script_file_path(batch_array)
with open(script_filename, "w") as fd:
fd.write(script)
os.chmod(
script_filename, stat.S_IWUSR | stat.S_IRUSR | stat.S_IRGRP
| stat.S_IROTH | stat.S_IXUSR | stat.S_IXGRP | stat.S_IXOTH)
priority = min(0, max(-1023, int(batch.priority * 1023 / 100 - 1023)))
try:
memory_limit = int(batch.memory_limit) / 1000
except:
memory_limit = 2
job_id = bputilities.run_on_tgt_os(
script=script,
group_name=batch.project,
job_name=batch_array.get_job_name(),
queue_name=batch.queue,
priority=priority,
memory=memory_limit,
cwd=cwd,
output=batch_array_text_file_path(batch_array),
err_output=batch_array_err_file_path(batch_array),
task_range=slice(TASK_ID_START, TASK_ID_START + len(bats)))
tasks = []
with bpcursor() as cursor:
job = BPJob.create(cursor, batch_array, job_id)
for bat in bats:
task = BPJobTask.create(cursor, job, bat)
tasks.append(task)
BPJobTaskStatus.create(cursor, task, JS_SUBMITTED)
return tasks
