def update_dmfilestat_diskusage(resultpk):
'''
Task to update DMFileStat.diskspace for all associated with this resultpk
This task is launched at the end of pipeline execution.
NOTE: This can be a long-lived task
'''
logid = {'logid':"%s" % ('tasks')}
try:
result = Results.objects.get(pk=resultpk)
search_dirs = [result.get_report_dir(), result.experiment.expDir]
cached_file_list = dm_utils.get_walk_filelist(search_dirs, list_dir=result.get_report_dir(), save_list=True)
for dmtype in FILESET_TYPES:
dmfilestat = result.get_filestat(dmtype)
dmfilestat_utils.update_diskspace(dmfilestat, cached=cached_file_list)
except SoftTimeLimitExceeded:
logger.warn("Time exceeded update_diskusage for (%d) %s" % (resultpk,result.resultsName), extra = logid)
except:
raise
try:
disk_total = 0
for dmfilestat in [ result.get_filestat(dmtype) for dmtype in FILESET_TYPES]:
if dmfilestat.dmfileset.type == dmactions_types.SIG:
dmfilestat.result.experiment.diskusage = dmfilestat.diskspace if dmfilestat.diskspace != None else 0
dmfilestat.result.experiment.save()
else:
partial = dmfilestat.diskspace
disk_total += int(partial) if partial != None else 0
result.diskusage = disk_total
result.save()
# See dmaction._update_related_objects() which also updates Exp & Results diskusage fields
except:
logger.error(traceback.format_exc(), extra = logid)
raise
def update_dmfilestats_diskspace(dmfilestat):
''' Task to update DMFileStat.diskspace '''
search_dirs = [dmfilestat.result.get_report_dir(), dmfilestat.result.experiment.expDir]
try:
cached_file_list = dm_utils.get_walk_filelist(search_dirs, list_dir=dmfilestat.result.get_report_dir())
dmfilestat_utils.update_diskspace(dmfilestat, cached=cached_file_list)
except:
logger.exception(traceback.format_exc(), extra = logid)
def update_dmfilestats_diskspace(dmfilestat):
''' Task to update DMFileStat.diskspace '''
search_dirs = [
dmfilestat.result.get_report_dir(), dmfilestat.result.experiment.expDir
]
try:
cached_file_list = dm_utils.get_walk_filelist(
search_dirs, list_dir=dmfilestat.result.get_report_dir())
dmfilestat_utils.update_diskspace(dmfilestat, cached=cached_file_list)
except:
logger.exception(traceback.format_exc(), extra=logid)
def backfill_dmfilestats_diskspace():
''' Backfill records with DMFileStat.diskspace = None, one at a time
These could be older data sets or new ones where update_diskusage task failed
'''
dmfilestats = DMFileStat.objects.filter(diskspace=None, action_state='L', files_in_use='').order_by('-created')
if dmfilestats.count() > 0:
dmfilestat = dmfilestats[0]
search_dirs = [dmfilestat.result.get_report_dir(), dmfilestat.result.experiment.expDir]
try:
cached_file_list = dm_utils.get_walk_filelist(search_dirs, list_dir=dmfilestat.result.get_report_dir(), save_list=True)
dmfilestat_utils.update_diskspace(dmfilestat, cached=cached_file_list)
except:
logger.error(traceback.format_exc(), extra = logid)
raise
def main(cl_args):
'''Main function'''
for dmtype in FILESET_TYPES:
sys.stdout.write("%s\n" % dmtype)
stats = models.DMFileStat.objects.filter(dmfileset__type=dmtype) \
.filter(action_state__in=['L', 'S', 'N', 'A']) \
.filter(diskspace=None) \
.order_by('created')
num_datasets = 0
for index, item in enumerate([stat for stat in stats if stat.diskspace is None]):
if index == 0:
sys.stdout.write("Missing disk space:\n")
num_datasets = index
if cl_args.repair:
sys.stdout.write("\t%s..." % (item.result.resultsName))
try:
if DEBUG:
sys.stdout.write("\nWe would run update_diskspace() here, if not in debug mode")
else:
sys.stdout.write("%d\n" % update_diskspace(item))
except:
print traceback.format_exc()
else:
sys.stdout.write("\t%s\n" % item.result.resultsName)
if num_datasets == 0:
sys.stdout.write("...No datasets found\n")
sys.exit(0)
def main(cl_args):
"""Main function"""
for dmtype in FILESET_TYPES:
sys.stdout.write("%s\n" % dmtype)
stats = (models.DMFileStat.objects.filter(
dmfileset__type=dmtype).filter(
action_state__in=["L", "S", "N", "A"]).filter(
diskspace=None).order_by("created"))
num_datasets = 0
for index, item in enumerate(
[stat for stat in stats if stat.diskspace is None]):
if index == 0:
sys.stdout.write("Missing disk space:\n")
num_datasets = index
if cl_args.repair:
sys.stdout.write("\t%s..." % (item.result.resultsName))
try:
if DEBUG:
sys.stdout.write(
"\nWe would run update_diskspace() here, if not in debug mode"
)
else:
sys.stdout.write("%d\n" % update_diskspace(item))
except Exception:
print(traceback.format_exc())
else:
sys.stdout.write("\t%s\n" % item.result.resultsName)
if num_datasets == 0:
sys.stdout.write("...No datasets found\n")
sys.exit(0)
def update_dmfilestat_diskusage(resultpk):
"""
Task to update DMFileStat.diskspace for all associated with this resultpk
This task is launched at the end of pipeline execution.
NOTE: This can be a long-lived task
"""
logid = {"logid": "%s" % ("tasks")}
try:
result = Results.objects.get(pk=resultpk)
search_dirs = [result.get_report_dir(), result.experiment.expDir]
cached_file_list = dm_utils.get_walk_filelist(
search_dirs, list_dir=result.get_report_dir(), save_list=True)
for dmtype in FILESET_TYPES:
dmfilestat = result.get_filestat(dmtype)
dmfilestat_utils.update_diskspace(dmfilestat,
cached=cached_file_list)
except SoftTimeLimitExceeded:
logger.warn(
"Time exceeded update_diskusage for (%d) %s" %
(resultpk, result.resultsName),
extra=logid,
)
except:
raise
try:
disk_total = 0
for dmfilestat in [
result.get_filestat(dmtype) for dmtype in FILESET_TYPES
]:
if dmfilestat.dmfileset.type == dmactions_types.SIG:
dmfilestat.result.experiment.diskusage = (
dmfilestat.diskspace
if dmfilestat.diskspace != None else 0)
dmfilestat.result.experiment.save()
else:
partial = dmfilestat.diskspace
disk_total += int(partial) if partial != None else 0
result.diskusage = disk_total
result.save()
# See dmaction._update_diskspace_and_diskusage() which also updates Exp & Results diskusage fields
except Exception:
logger.error(traceback.format_exc(), extra=logid)
raise
def backfill_dmfilestats_diskspace():
''' Backfill records with DMFileStat.diskspace = None, one at a time
These could be older data sets or new ones where update_diskusage task failed
'''
dmfilestats = DMFileStat.objects.filter(
diskspace=None, action_state='L', files_in_use='').order_by('-created')
if dmfilestats.count() > 0:
dmfilestat = dmfilestats[0]
search_dirs = [
dmfilestat.result.get_report_dir(),
dmfilestat.result.experiment.expDir
]
try:
cached_file_list = dm_utils.get_walk_filelist(
search_dirs,
list_dir=dmfilestat.result.get_report_dir(),
save_list=True)
dmfilestat_utils.update_diskspace(dmfilestat,
cached=cached_file_list)
except:
logger.error(traceback.format_exc(), extra=logid)
raise
def _action_complete_update(user, user_comment, dmfilestat, action):
logger.debug("Function: %s()" % sys._getframe().f_code.co_name, extra = logid)
if action == ARCHIVE:
action_state = 'AD'
msg = "%0.1f MB %s archived to %s." % (dmfilestat.diskspace, dmfilestat.dmfileset.type, dmfilestat.archivepath)
elif action == DELETE:
action_state = 'DD'
msg = "%0.1f MB %s deleted." % (dmfilestat.diskspace, dmfilestat.dmfileset.type)
elif action == EXPORT:
action_state = 'L'
msg = "%0.1f MB %s exported to %s." % (dmfilestat.diskspace, dmfilestat.dmfileset.type, dmfilestat.archivepath)
dmfilestat.archivepath = None
dmfilestat.save()
elif action == TEST:
return
# Update disk usage values in dmfilestat objects after a delete or archive action. Should be zero or close.
# update diskusage on Experiment or Results object if data was deleted/moved
if action != EXPORT:
diskspace = update_diskspace(dmfilestat)
if dmfilestat.dmfileset.type == dmactions_types.SIG:
dmfilestat.result.experiment.diskusage = diskspace if diskspace != None else 0
dmfilestat.result.experiment.save()
else:
result = dmfilestat.result
disk_total = 0
mylist = [
result.get_filestat(dmactions_types.BASE).diskspace,
result.get_filestat(dmactions_types.OUT).diskspace,
result.get_filestat(dmactions_types.INTR).diskspace
]
for partial in mylist:
disk_total += int(partial) if partial != None else 0
result.diskusage = disk_total
result.save()
# See data/tasks.update_dmfilestat_diskusage() which also updates Exp & Results diskusage fields
_update_related_objects(user, user_comment, dmfilestat, action, msg, action_state)
def _update_diskspace_and_diskusage(dmfilestat):
logger.debug("Function: %s()" % sys._getframe().f_code.co_name, extra=logid)
diskspace = update_diskspace(dmfilestat)
# update diskusage on Experiment or Results object if data was deleted/moved
# See data/tasks.update_dmfilestat_diskusage() which also updates Exp & Results diskusage fields
if dmfilestat.dmfileset.type == dmactions_types.SIG:
dmfilestat.result.experiment.diskusage = diskspace if diskspace != None else 0
dmfilestat.result.experiment.save()
else:
result = dmfilestat.result
disk_total = 0
mylist = [
result.get_filestat(dmactions_types.BASE).diskspace,
result.get_filestat(dmactions_types.OUT).diskspace,
result.get_filestat(dmactions_types.INTR).diskspace
]
for partial in mylist:
disk_total += int(partial) if partial != None else 0
result.diskusage = disk_total
result.save()
def destination_validation(dmfilestat, backup_directory=None, manual_action=False):
'''
Tests to validate destination directory:
Does destination directory exist.
Is there sufficient disk space.
Write permissions.
'''
def _skipdiskspacecheck(directory):
'''
The hidden file .no_size_check should be placed into the root directory of the scratch drive mounted on the local
system for the tape drive system.
'''
if os.path.exists(os.path.join(directory, ".no_size_check")):
logger.info("%s: Exists: %s" %
(sys._getframe().f_code.co_name, os.path.join(directory, ".no_size_check")), extra=logid)
return True
else:
logger.info("%s: Not Found: %s" %
(sys._getframe().f_code.co_name, os.path.join(directory, ".no_size_check")), extra=logid)
return False
logger.debug("Function: %s()" % sys._getframe().f_code.co_name, extra=logid)
if backup_directory in [None, 'None', '']:
backup_directory = dmfilestat.dmfileset.backup_directory
# check for valid destination
try:
if backup_directory in [None, 'None', '', '/']:
raise DMExceptions.MediaNotSet(
"Backup media for %s is not configured. Please use Data Management Configuration page." % dmfilestat.dmfileset.type)
if not os.path.isdir(backup_directory):
raise DMExceptions.MediaNotAvailable(
"Backup media for %s is not a directory: %s" % (dmfilestat.dmfileset.type, backup_directory))
# check if destination is external filesystem
# ie, catch the error of writing to a mountpoint which is unmounted.
# Use the tool 'mountpoint' which returns 0 if its mountpoint and mounted
# If its a subdirectory to a mountpoint, then the isdir() test will fail when not mounted.
if not is_mounted(backup_directory):
raise DMExceptions.MediaNotAvailable(
"Backup media for %s is not mounted: %s" % (dmfilestat.dmfileset.type, backup_directory))
except Exception as e:
logger.error("%s" % e, extra=logid)
raise
# check for sufficient disk space (units: kilobytes)
if _skipdiskspacecheck(backup_directory):
logger.warn("%s - skipping destination disk space check" %
(sys._getframe().f_code.co_name), extra=logid)
pass
else:
if dmfilestat.diskspace is None:
diskspace = update_diskspace(dmfilestat)
else:
diskspace = dmfilestat.diskspace
try:
freespace = getSpaceMB(backup_directory)
pending = 0
if manual_action:
# add up all objects that will be processed before this one
exp_ids = []
for obj in DMFileStat.objects.filter(action_state__in=['AG', 'EG', 'SA', 'SE']):
try:
if obj.archivepath and not os.path.normpath(obj.archivepath).startswith(os.path.normpath(backup_directory)):
continue
elif not os.path.normpath(obj.dmfileset.backup_directory).startswith(os.path.normpath(backup_directory)):
continue
if obj.dmfileset.type == dmactions_types.SIG:
if obj.result.experiment_id not in exp_ids:
exp_ids.append(obj.result.experiment_id)
pending += obj.diskspace
else:
pending += obj.diskspace
except:
pass
logger.debug("Required %dMB, pending %dMB, free %dMB" %
(diskspace, pending, freespace), extra=logid)
if diskspace >= freespace:
raise DMExceptions.InsufficientDiskSpace(
"Not enough space to write files at %s (free=%dMB)" % (backup_directory, freespace))
elif diskspace >= (freespace - pending):
raise DMExceptions.InsufficientDiskSpace(
"Not enough space to write files at %s (free=%dMB, pending=%dMB)" % (backup_directory, freespace, pending))
except Exception as e:
logger.debug("%s" % str(e), extra=logid)
raise
# check for write permission.
cmd = ['sudo', '/opt/ion/iondb/bin/sudo_utils.py', 'check_write_permission', backup_directory]
try:
process = subprocess.Popen(cmd, stdout=subprocess.PIPE, stderr=subprocess.PIPE)
_, stderrdata = process.communicate()
except Exception as err:
logger.debug("Sub Process execution failed %s" % err, extra=logid)
if process.returncode:
raise DMExceptions.FilePermission(stderrdata)
# check for existence of source directory - needed first to create destination folder
# and, of course, as source of files to copy. but we check here instead of failing halfway
# thru the setup. This shows inconsistency between dmfilestat action_status and filesystem.
if dmfilestat.dmfileset.type == dmactions_types.SIG:
src_dir = dmfilestat.result.experiment.expDir
else:
src_dir = dmfilestat.result.get_report_dir()
if not os.path.exists(src_dir):
raise DMExceptions.SrcDirDoesNotExist(src_dir)
def destination_validation(dmfilestat, backup_directory=None, manual_action=False):
'''
Tests to validate destination directory:
Does destination directory exist.
Is there sufficient disk space.
Write permissions.
'''
def _skipdiskspacecheck(directory):
'''
The hidden file .no_size_check should be placed into the root directory of the scratch drive mounted on the local
system for the tape drive system.
'''
if os.path.exists(os.path.join(directory, ".no_size_check")):
logger.info("%s: Exists: %s" % (sys._getframe().f_code.co_name, os.path.join(directory, ".no_size_check")))
return True
else:
logger.info("%s: Not Found: %s" % (sys._getframe().f_code.co_name, os.path.join(directory, ".no_size_check")))
return False
logger.debug("Function: %s()" % sys._getframe().f_code.co_name, extra = logid)
if backup_directory in [None, 'None', '']:
backup_directory = dmfilestat.dmfileset.backup_directory
# check for valid destination
try:
if backup_directory in [None, 'None', '', '/']:
raise DMExceptions.MediaNotSet("Backup media for %s is not configured. Please use Data Management Configuration page." % dmfilestat.dmfileset.type)
if not os.path.isdir(backup_directory):
raise DMExceptions.MediaNotAvailable("Backup media for %s is not a directory: %s" % (dmfilestat.dmfileset.type, backup_directory))
# check if destination is external filesystem
# ie, catch the error of writing to a mountpoint which is unmounted.
# Use the tool 'mountpoint' which returns 0 if its mountpoint and mounted
# If its a subdirectory to a mountpoint, then the isdir() test will fail when not mounted.
if not is_mounted(backup_directory):
raise DMExceptions.MediaNotAvailable("Backup media for %s is not mounted: %s" % (dmfilestat.dmfileset.type, backup_directory))
except Exception as e:
logger.error("%s" % e, extra = logid)
raise
# check for sufficient disk space (units: kilobytes)
if _skipdiskspacecheck(backup_directory):
logger.warn("%s - skipping destination disk space check" % (sys._getframe().f_code.co_name), extra = logid)
pass
else:
if dmfilestat.diskspace is None:
diskspace = update_diskspace(dmfilestat)
else:
diskspace = dmfilestat.diskspace
try:
freespace = getSpaceMB(backup_directory)
pending = 0
if manual_action:
# add up all objects that will be processed before this one
exp_ids = []
for obj in DMFileStat.objects.filter(action_state__in=['AG', 'EG', 'SA', 'SE']):
try:
if obj.archivepath and not os.path.normpath(obj.archivepath).startswith(os.path.normpath(backup_directory)):
continue
elif not os.path.normpath(obj.dmfileset.backup_directory).startswith(os.path.normpath(backup_directory)):
continue
if obj.dmfileset.type == dmactions_types.SIG:
if obj.result.experiment_id not in exp_ids:
exp_ids.append(obj.result.experiment_id)
pending += obj.diskspace
else:
pending += obj.diskspace
except:
pass
logger.debug("Required %dMB, pending %dMB, free %dMB" % (diskspace, pending, freespace), extra = logid)
if diskspace >= freespace:
raise DMExceptions.InsufficientDiskSpace("Not enough space to write files at %s (free=%dMB)" % (backup_directory, freespace))
elif diskspace >= (freespace - pending):
raise DMExceptions.InsufficientDiskSpace("Not enough space to write files at %s (free=%dMB, pending=%dMB)" % (backup_directory, freespace, pending))
except Exception as e:
logger.debug("%s" % str(e), extra = logid)
raise
# check for write permission. NOTE: this function is called by apache2 user while the action function
# will be executed within a celery task which takes the uid/gid of celeryd process - in our case that is currently root.
# This test is too restrictive.
try:
foo = tempfile.NamedTemporaryFile(dir=backup_directory)
foo.close()
except Exception as e:
if e.errno in [errno.EPERM, errno.EACCES]: # Operation not permitted
errmsg = "Insufficient write permission in %s" % backup_directory
else:
errmsg = e
logger.error(errmsg, extra = logid)
raise DMExceptions.FilePermission(errmsg)
# check for existence of source directory - needed first to create destination folder
# and, of course, as source of files to copy. but we check here instead of failing halfway
# thru the setup. This shows inconsistency between dmfilestat action_status and filesystem.
if dmfilestat.dmfileset.type == dmactions_types.SIG:
src_dir = dmfilestat.result.experiment.expDir
else:
src_dir = dmfilestat.result.get_report_dir()
if not os.path.exists(src_dir):
raise DMExceptions.SrcDirDoesNotExist(src_dir)
