def lsf_run_until_done(to_run_dict, logfile, queue, bsub_flags, jobname_base,
num_batches, MAX_RETRY):
from run_safe import unfinished_cmds
cmds = unfinished_cmds(to_run_dict)
retries = 0
last_cmds = []
while len(cmds) > 0:
print >> sys.stderr, '%s: %s cmds to run in %s batches on queue %s, logs in %s' % (
jobname_base, len(cmds), num_batches, queue, logfile)
#code to halt execution on recurrent errors
if set(last_cmds) == set(cmds):
if retries > MAX_RETRY:
errstr = 'maximum number of retry attempts (%s) exceeded with identical jobs lists. Check logs (%s) for recurrent errors' % (
MAX_RETRY, logfile)
raise IOError, errstr
else:
retries += 1
last_cmds = cmds
jobids, namedict = lsf_jobs_submit(cmds,
logfile,
queue,
bsub_flags,
jobname_base=jobname_base,
num_batches=num_batches)
time.sleep(20)
lsf_wait_for_jobs(jobids, logfile, namedict=namedict)
cmds = unfinished_cmds(to_run_dict)
print >> sys.stderr, 'DONE\n'
def run_until_done(to_run_dict,jobname_base,scriptdir, runtime,mem, num_batches, partition='general' ,force_source=False,MAX_RETRY=MAX_RETRY,**kwargs):
'''given to-run dictionary as populated by run_safe.add_cmd (see run_safe.py in py_util) and scheduling parameters
submits jobs that have not yet completed per run_safe .done files until all jobs finish or until identical job lists are submitted MAX_RETRY times
see jobs_submit and wait_for_jobs in this module for more details
kwargs go to jobs_submit; see jobs_submit and slurm_script for handling of additional arguments
'''
from run_safe import unfinished_cmds
cmds = unfinished_cmds(to_run_dict)
retries = 0
last_cmds = []
while len(cmds) > 0:
print >> sys.stderr, '%s: %s cmds to run in %s batches on queue %s, logs in %s' % (jobname_base,len(cmds),num_batches,partition,scriptdir)
#code to halt execution on recurrent errors
if set(last_cmds) == set(cmds):
if retries > MAX_RETRY:
errstr = 'maximum number of retry attempts (%s) exceeded with identical jobs lists. Check logs (%s) for recurrent errors' % (MAX_RETRY,scriptdir)
raise IOError, errstr
else:
retries += 1
last_cmds = cmds
jobsdict = jobs_submit(cmds,jobname_base,scriptdir, runtime,mem, num_batches,partition=partition ,force_source=force_source, **kwargs)
time.sleep(20)
wait_for_jobs(jobsdict,restart_partition=partition,sleeptime = 20)
time.sleep(20)
cmds = unfinished_cmds(to_run_dict)
print >> sys.stderr, 'DONE\n'
def run_until_done(to_run_dict,jobname_base,scriptdir, runtime,mem, num_batches, partition='general' ,force_source=False,MAX_RETRY=MAX_RETRY,**kwargs):
'''given to-run dictionary as populated by run_safe.add_cmd (see run_safe.py in py_util) and scheduling parameters
submits jobs that have not yet completed per run_safe .done files until all jobs finish or until identical job lists are submitted MAX_RETRY times
see jobs_submit and wait_for_jobs in this module for more details
kwargs go to jobs_submit; see jobs_submit and slurm_script for handling of additional arguments
'''
from run_safe import unfinished_cmds
cmds = unfinished_cmds(to_run_dict)
retries = 0
last_cmds = []
while len(cmds) > 0:
print >> sys.stderr, '%s: %s cmds to run in %s batches on queue %s, logs in %s' % (jobname_base,len(cmds),num_batches,partition,scriptdir)
#code to halt execution on recurrent errors
if set(last_cmds) == set(cmds):
if retries > MAX_RETRY:
errstr = 'maximum number of retry attempts (%s) exceeded with identical jobs lists. Check logs (%s) for recurrent errors' % (MAX_RETRY,scriptdir)
raise IOError, errstr
else:
retries += 1
last_cmds = cmds
jobsdict = jobs_submit(cmds,jobname_base,scriptdir, runtime,mem, num_batches,partition=partition ,force_source=force_source, **kwargs)
time.sleep(20)
wait_for_jobs(jobsdict,restart_partition=partition,sleeptime = 20)
time.sleep(20)
cmds = unfinished_cmds(to_run_dict)
print >> sys.stderr, 'DONE\n'
def lsf_run_until_done(to_run_dict,logfile,queue,bsub_flags,jobname_base,num_batches,MAX_RETRY):
from run_safe import unfinished_cmds
cmds = unfinished_cmds(to_run_dict)
retries = 0
last_cmds = []
while len(cmds) > 0:
print >> sys.stderr, '%s: %s cmds to run in %s batches on queue %s, logs in %s' % (jobname_base,len(cmds),num_batches,queue,logfile)
#code to halt execution on recurrent errors
if set(last_cmds) == set(cmds):
if retries > MAX_RETRY:
errstr = 'maximum number of retry attempts (%s) exceeded with identical jobs lists. Check logs (%s) for recurrent errors' % (MAX_RETRY,logfile)
raise IOError, errstr
else:
retries += 1
last_cmds = cmds
jobids,namedict = lsf_jobs_submit(cmds,logfile,queue,bsub_flags,jobname_base=jobname_base,num_batches=num_batches)
time.sleep(20)
lsf_wait_for_jobs(jobids,logfile,namedict=namedict)
cmds = unfinished_cmds(to_run_dict)
print >> sys.stderr, 'DONE\n'
