def get_scratch_dir():
scratch = ''
# Get scratch directory path.
if 'TMPDIR' in os.environ:
scratch = os.environ['TMPDIR']
elif 'SCRATCH' in os.environ:
scratch = os.environ['SCRATCH']
else:
scratch = '/scratch/%s/%s' % (get_experiment(), get_user())
if not larbatch_posix.isdir(scratch) or not larbatch_posix.access(
scratch, os.W_OK):
scratch = '/%s/data/users/%s' % (get_experiment(), get_user())
# Checkout.
if scratch == '':
raise RuntimeError('No scratch directory specified.')
if not larbatch_posix.isdir(scratch) or not larbatch_posix.access(
scratch, os.W_OK):
raise RuntimeError(
'Scratch directory %s does not exist or is not writeable.' %
scratch)
return scratch
def active_projects(defname=''):
result = set()
# Get project name stem.
s = samweb()
prjstem = ''
if defname != '':
prjstem = '%s_' % s.makeProjectName(defname).rsplit('_', 1)[0]
# Dump station
url = '%s/dumpStation?station=%s' % (s.get_baseurl(), get_experiment())
furl = urlrequest.urlopen(url)
# Parse response.
for line in furl.readlines():
words = line.split()
if len(words) > 5:
prjname = convert_str(words[0])
if prjstem == '' or prjname.startswith(prjstem):
result.add(prjname)
# Done.
return result
def makeDummyDef(defname):
if not defExists(defname):
# Make dummy definition.
print('Making dummy dataset definition %s' % defname)
test_kca()
samweb().createDefinition(defname, 'file_id 0', user=get_user(), group=get_experiment())
def samweb():
global samweb_obj
if samweb_obj == None:
samweb_obj = samweb_cli.SAMWebClient(experiment=get_experiment())
os.environ['SSL_CERT_DIR'] = '/etc/grid-security/certificates'
return samweb_obj
def active_projects2(defname=''):
result = set()
# Get project name stem.
s = samweb()
prjstem = ''
if defname != '':
prjstem = '%s_' % s.makeProjectName(defname).rsplit('_', 1)[0]
# Query a list of projects started within the last 72 hours.
dt = datetime.timedelta(3, 0)
tmin = datetime.datetime.utcnow() - dt
tminstr = tmin.strftime('%Y-%m-%dT%H:%M:%S')
prjnames = s.listProjects(started_after=tminstr)
# Loop over projects to check end times.
for prjname in prjnames:
if prjstem == '' or prjname.startswith(prjstem):
# This project is a candidate for inclusion in result.
# Check end time.
age = 0
prjurl = s.findProject(project=prjname, station=get_experiment())
if prjurl != '':
prjsum = s.projectSummary(prjurl)
if 'project_end_time' in prjsum:
tendstr = prjsum['project_end_time']
if len(tendstr) >= 19:
try:
tend = datetime.datetime.strptime(
tendstr[:19], '%Y-%m-%dT%H:%M:%S')
tage = datetime.datetime.utcnow() - tend
age = tage.total_seconds()
except:
pass
# Keep this project if there is no end time.
if age == 0:
result.add(prjname)
# Done.
return result
def create_limited_dataset(defname, run, subruns):
if len(subruns) == 0:
return ''
# Construct comma-separated list of run-subrun pairs in a form that is
# acceptable as sam dimension constraint.
run_subrun_dim = ''
for subrun in subruns:
if run_subrun_dim != '':
run_subrun_dim += ','
run_subrun_dim += "%d.%d" % (run, subrun)
# Construct dimension including run and subrun constraints.
dim = "defname: %s and run_number %s" % (defname, run_subrun_dim)
# Test the new dimension.
nfiles = samweb().countFiles(dimensions=dim)
if nfiles == 0:
return ''
# Make sure we have a kca certificate.
test_kca()
# Construct a new unique definition name.
newdefname = defname + '_' + str(uuid.uuid4())
# Create definition.
samweb().createDefinition(newdefname,
dim,
user=get_user(),
group=get_experiment())
# Done (return definition name).
return newdefname
def makeFileListDefinition(list_or_dim):
# Make sure we have a kca certificate.
test_kca()
# Make file list dimension.
flist = []
if type(list_or_dim) == type([]) or type(list_or_dim) == type(set()):
flist = list_or_dim
print('Making file list definition from %s with %d elements.' %
(type(list_or_dim), len(list_or_dim)))
else:
flist = listFiles(list_or_dim)
print('Making file list definition using dimension "%s"' % list_or_dim)
listdim = ''
for filename in flist:
if listdim == '':
listdim = 'file_name %s' % filename
else:
listdim += ', %s' % filename
if listdim == '':
listdim = 'file_id 0'
# Maybe construct a new unique definition name.
defname = get_user() + '_filelist_' + str(uuid.uuid4())
# Create definition.
samweb().createDefinition(defname,
listdim,
user=get_user(),
group=get_experiment())
# Done.
return defname
def active_projects(defname=''):
result = set()
# Get project name stem.
s = samweb()
prjstem = ''
if defname != '':
prjstem = '%s_' % s.makeProjectName(defname).rsplit('_', 1)[0]
# Dump station
url = '%s/dumpStation?station=%s' % (s.get_baseurl(), get_experiment())
buffer = StringIO.StringIO()
c = pycurl.Curl()
c.setopt(c.URL, url)
c.setopt(c.USERPWD, 'uboone:argon!')
c.setopt(c.FOLLOWLOCATION, True)
c.setopt(c.WRITEFUNCTION, buffer.write)
c.perform()
c.close()
# Parse response.
buffer.seek(0)
for line in buffer.readlines():
words = line.split()
if len(words) > 0 and words[0] == 'project':
prjname = words[1].split('(')[0]
if prjstem == '' or prjname.startswith(prjstem):
result.add(prjname)
# Done.
return result
def default_jobsub_submit_options():
opt = '--append_condor_requirements=\'(TARGET.HAS_CVMFS_%s_opensciencegrid_org==true)\'' % get_experiment(
)
return opt
def make_active_project_dataset(defname, dropboxwait, active_defname,
wait_defname):
s = samweb()
test_kca()
# Get list of active projects.
prjs = active_projects(defname) | active_projects2(defname)
# Make sam dimension.
dim = ''
for prj in prjs:
if dim == '':
dim = 'snapshot_for_project_name %s' % prj
else:
dim += ',%s' % prj
# If there were no matching projects, make up some legal dimension that won't
# match any files.
if dim == '':
dim = 'file_id 0'
# Create or update active_defname.
if defExists(active_defname):
print('Updating dataset definition %s' % active_defname)
s.deleteDefinition(active_defname)
else:
print('Creating dataset definition %s' % active_defname)
s.createDefinition(active_defname,
dim,
user=get_user(),
group=get_experiment())
# If the dropbox waiting interval is nonzero, create a dataset for
# dropbox waiting files.
dim = ''
if dropboxwait > 0.:
# Convert the waiting interval to a datetime.timedelta object.
dt = datetime.timedelta(int(dropboxwait), int(dropboxwait % 1 * 86400))
# Get the earliest allowed time.
tmin = datetime.datetime.utcnow() - dt
# Format time in a form acceptable to sam.
tminstr = tmin.strftime('%Y-%m-%dT%H:%M:%S')
# Append sam dimension.
dim = "isparentof: (create_date > '%s' and availability: virtual)" % tminstr
else:
# Otherwise make dummy dataset.
dim = 'file_id 0'
# Create or update active_defname.
if defExists(wait_defname):
print('Updating dataset definition %s' % wait_defname)
s.deleteDefinition(wait_defname)
else:
print('Creating dataset definition %s' % wait_defname)
s.createDefinition(wait_defname,
dim,
user=get_user(),
group=get_experiment())
def start_project(defname, default_prjname, max_files, force_snapshot,
filelistdef):
# Check project name.
s = samweb()
prjname = default_prjname
if prjname == '':
prjname = s.makeProjectName(defname)
print('Starting project %s' % prjname)
# Make sure we have a certificate.
test_kca()
# Figure out how many files are in the input dataset.
nf = 0
if filelistdef:
files = listFiles('defname: %s' % defname)
nf = len(files)
else:
nf = s.countFiles('defname: %s' % defname)
print('Input dataset has %d files.' % nf)
if nf == 0:
return 1
# Make limited dataset?
if max_files > 0 and nf > max_files:
limitdef = '%s_limit_%d' % (prjname, max_files)
# Figure out whether limitdef already exists.
if defExists(limitdef) and not filelistdef:
print('Using already created limited dataset definition %s.' %
limitdef)
else:
dim = 'defname: %s with limit %d' % (defname, max_files)
if filelistdef:
limitdef = makeFileListDefinition(dim)
else:
print('Creating limited dataset definition %s.' % limitdef)
s.createDefinition(limitdef,
dim,
user=get_user(),
group=get_experiment())
defname = limitdef
nf = max_files
elif filelistdef:
dim = 'defname: %s' % defname
defname = makeFileListDefinition(dim)
# Force snapshot?
if force_snapshot:
print('Forcing snapthot.')
defname = '%s:force' % defname
# Start the project.
print('Starting project %s.' % prjname)
s.startProject(prjname,
defname=defname,
station=get_experiment(),
group=get_experiment(),
user=get_user())
# Done.
return 0
