def run(self, args):
"""Run 'connect' subcommand."""
optparser = ConnectOptions(go_args=args,
envvar_prefix=self.envvar_prefix,
usage=self.usage_txt)
try:
if len(optparser.args) > 1:
label = optparser.args[1]
else:
self.report_error("No label provided.")
print "Connecting to HOD cluster with label '%s'..." % label
try:
jobid = cluster_jobid(label)
env_script = cluster_env_file(label)
except ValueError as err:
self.report_error(err)
print "Job ID found: %s" % jobid
pbs = rm_pbs.Pbs(optparser)
jobs = pbs.state()
pbsjobs = [job for job in jobs if job.jobid == jobid]
if len(pbsjobs) == 0:
self.report_error("Job with job ID '%s' not found by pbs.",
jobid)
elif len(pbsjobs) > 1:
self.report_error("Multiple jobs found with job ID '%s': %s",
jobid, pbsjobs)
pbsjob = pbsjobs[0]
if pbsjob.state == ['Q', 'H']:
# This should never happen since the hod.d/<jobid>/env file is
# written on cluster startup. Maybe someone hacked the dirs.
self.report_error(
"Cannot connect to cluster with job ID '%s' yet. It is still queued.",
jobid)
else:
print "HOD cluster '%s' @ job ID %s appears to be running..." % (
label, jobid)
print "Setting up SSH connection to %s..." % pbsjob.hosts
# -i: interactive non-login shell
cmd = [
'ssh', '-t', pbsjob.hosts, 'exec', 'bash', '--rcfile',
env_script, '-i'
]
self.log.info("Logging in using command: %s", ' '.join(cmd))
os.execvp('/usr/bin/ssh', cmd)
except StandardError as err:
self._log_and_raise(err)
return 0
def run(self, args):
"""Run 'connect' subcommand."""
optparser = ConnectOptions(go_args=args, envvar_prefix=self.envvar_prefix, usage=self.usage_txt)
try:
if len(optparser.args) > 1:
label = optparser.args[1]
else:
_log.error("No label provided.")
sys.exit(1)
print "Connecting to HOD cluster with label '%s'..." % label
try:
jobid = cluster_jobid(label)
env_script = cluster_env_file(label)
except ValueError as err:
_log.error(err)
sys.exit(1)
print "Job ID found: %s" % jobid
pbs = rm_pbs.Pbs(optparser)
jobs = pbs.state()
pbsjobs = [job for job in jobs if job.jobid == jobid]
if len(pbsjobs) == 0:
_log.error("Job with job ID '%s' not found by pbs.", jobid)
sys.exit(1)
elif len(pbsjobs) > 1:
_log.error("Multiple jobs found with job ID '%s': %s", jobid, pbsjobs)
sys.exit(1)
pbsjob = pbsjobs[0]
if pbsjob.state == ['Q', 'H']:
# This should never happen since the hod.d/<jobid>/env file is
# written on cluster startup. Maybe someone hacked the dirs.
_log.error("Cannot connect to cluster with job ID '%s' yet. It is still queued.", jobid)
sys.exit(1)
else:
print "HOD cluster '%s' @ job ID %s appears to be running..." % (label, jobid)
print "Setting up SSH connection to %s..." % pbsjob.hosts
# -i: interactive non-login shell
cmd = ['ssh', '-t', pbsjob.hosts, 'exec', 'bash', '--rcfile', env_script, '-i']
_log.info("Logging in using command: %s", ' '.join(cmd))
os.execvp('/usr/bin/ssh', cmd)
return 0 # pragma: no cover
except StandardError as err:
fancylogger.setLogFormat(fancylogger.TEST_LOGGING_FORMAT)
fancylogger.logToScreen(enable=True)
_log.raiseException(err)
def run(self, args):
"""Run 'destroy' subcommand."""
optparser = DestroyOptions(go_args=args, envvar_prefix=self.envvar_prefix, usage=self.usage_txt)
try:
label, jobid = None, None
if len(optparser.args) > 1:
label = optparser.args[1]
print "Destroying HOD cluster with label '%s'..." % label
else:
_log.error("No label provided.")
sys.exit(1)
try:
jobid = cluster_jobid(label)
print "Job ID: %s" % jobid
except ValueError as err:
_log.error(err)
sys.exit(1)
# try to figure out job state
job_state = None
pbs = rm_pbs.Pbs(optparser)
jobs = pbs.state()
pbsjobs = [job for job in jobs if job.jobid == jobid]
_log.debug("Matching jobs for job ID '%s': %s", jobid, pbsjobs)
if len(pbsjobs) == 1:
job_state = pbsjobs[0].state
print "Job status: %s" % job_state
elif len(pbsjobs) == 0:
print "(job no longer found)"
else:
_log.error("Multiple jobs found with job ID '%s': %s", jobid, pbsjobs)
sys.exit(1)
# request confirmation is case the job is currently running
if job_state == 'R':
resp = raw_input("Confirm destroying the *running* HOD cluster with label '%s'? [y/n]: " % label)
if resp != 'y':
print "(destruction aborted)"
return
elif job_state in ['C', 'E']:
print "(job has already ended/completed)"
job_state = None
print "\nStarting actual destruction of HOD cluster with label '%s'...\n" % label
# actually destroy HOD cluster by deleting job and removing cluster info dir and local work dir
if job_state is not None:
# if job was not successfully deleted, pbs.remove will print an error message
if pbs.remove(jobid):
print "Job with ID %s deleted." % jobid
rm_cluster_localworkdir(label)
if cluster_info_exists(label):
rm_cluster_info(label)
print "\nHOD cluster with label '%s' (job ID: %s) destroyed." % (label, jobid)
except StandardError as err:
fancylogger.setLogFormat(fancylogger.TEST_LOGGING_FORMAT)
fancylogger.logToScreen(enable=True)
_log.raiseException(err)
def run(self, args):
"""Run 'destroy' subcommand."""
optparser = DestroyOptions(go_args=args,
envvar_prefix=self.envvar_prefix,
usage=self.usage_txt)
try:
label, jobid = None, None
if len(optparser.args) > 1:
label = optparser.args[1]
print "Destroying HOD cluster with label '%s'..." % label
else:
self.report_error("No label provided.")
try:
jobid = cluster_jobid(label)
print "Job ID: %s" % jobid
except ValueError as err:
self.report_error(err)
# try to figure out job state
job_state = None
pbs = rm_pbs.Pbs(optparser)
jobs = pbs.state()
pbsjobs = [job for job in jobs if job.jobid == jobid]
self.log.debug("Matching jobs for job ID '%s': %s", jobid, pbsjobs)
if len(pbsjobs) == 1:
job_state = pbsjobs[0].state
print "Job status: %s" % job_state
elif len(pbsjobs) == 0:
print "(job no longer found)"
else:
self.report_error("Multiple jobs found with job ID '%s': %s",
jobid, pbsjobs)
# request confirmation is case the job is currently running
if job_state == 'R':
resp = raw_input(
"Confirm destroying the *running* HOD cluster with label '%s'? [y/n]: "
% label)
if resp != 'y':
print "(destruction aborted)"
return
elif job_state in ['C', 'E']:
print "(job has already ended/completed)"
job_state = None
print "\nStarting actual destruction of HOD cluster with label '%s'...\n" % label
# actually destroy HOD cluster by deleting job and removing cluster info dir and local work dir
if job_state is not None:
# if job was not successfully deleted, pbs.remove will print an error message
if pbs.remove(jobid):
print "Job with ID %s deleted." % jobid
rm_cluster_localworkdir(label)
if cluster_info_exists(label):
rm_cluster_info(label)
print "\nHOD cluster with label '%s' (job ID: %s) destroyed." % (
label, jobid)
except StandardError as err:
self._log_and_raise(err)
return 0