def main():
"""Main script."""
options = {
'nagios-check-interval-threshold': NAGIOS_CHECK_INTERVAL_THRESHOLD,
'mail-report': ('mail a report to the hpc-admin list with job list for gracing or inactive users',
None, 'store_true', False),
'access_token': ('OAuth2 token to access the account page REST API', None, 'store', None),
'account_page_url': ('URL of the account page where we can find the REST API', None, 'store', None)
}
opts = ExtendedSimpleOption(options)
try:
now = datetime.datetime.utcnow()
timestamp = now - datetime.timedelta(days=1)
client = AccountpageClient(token=opts.options.access_token, url=opts.options.account_page_url + "/api/")
active_users, inactive_users = client.get_accounts()
grace_users = []
for a in active_users:
try:
if a.expiry_date and datetime.datetime.strptime(a.expiry_date, "%Y-%m-%d") - now < datetime.timedelta(days=7):
grace_users.append(a)
except AttributeError as err:
logger.debug("Account %s does not have expiry date", a.vsc_id)
pbs_query = PBSQuery()
t = time.ctime()
jobs = pbs_query.getjobs() # we just get them all
removed_queued = remove_queued_jobs(jobs, grace_users, inactive_users)
removed_running = remove_running_jobs(jobs, inactive_users)
if opts.options.mail_report and not opts.options.dry_run:
if len(removed_queued) > 0 or len(removed_running) > 0:
mail_report(t, removed_queued, removed_running)
except Exception, err:
logger.exception("critical exception caught: %s" % (err))
opts.critical("Script failed in a horrible way")
sys.exit(NAGIOS_EXIT_CRITICAL)
def main():
"""
Main script. Sets the hard limit for either user or VO.
"""
options = {
'storage': ("The storage system's name", None, 'store', None),
'fileset': ("The fileset where you want to adjust the quota", None, 'store', None),
'user': ('process users', None, 'store', None),
'vo': ('process vos', None, 'store', None),
'size': ('the target quota (in KiB)', int, 'store', None),
'original': ('show the original quota values', None, 'store_true', False),
'access_token': ('OAuth2 token to access the account page REST API', None, 'store', None),
}
opts = SimpleOption(options)
client = AccountpageClient(token=opts.options.access_token)
fileset = opts.options.fileset
storage = opts.options.storage
if not opts.options.size:
print "size "
sys.exit()
# TODO: could use some love in allowing a unit to be appended and converting to KiB prior to uploading
size = opts.options.size
if opts.options.user:
# quota/user/%(vsc_id)s/storage/%(storage)s/fileset/%(fileset)s/size/$
vsc_id = opts.options.user
original = client.account[vsc_id].quota
upload = client.quota.user[vsc_id].storage[storage].fileset[fileset].size
if opts.options.vo:
vsc_id = opts.options.vo
original = client.vo[vsc_id].quota
upload = client.quota.vo[vsc_id].storage[storage].fileset[fileset].size
current = original.get()
if current[0] in (200,):
if opts.options.original:
print("Original values: %s" % (current[1],))
else:
print("Error, could not get original quota values for the given parameters")
print("Issue: %s" % (current[1],))
sys.exit(-1)
result = upload.put(body={"hard": size})
if result[0] in (200,):
print("Request OK.")
print("New values: %s" % (result[1],))
else:
print("Request failed")
print("Issue: %s" % (result[1],))
def main():
"""
Main script.
- build the filter
- fetches the users
- process the users
- write the new timestamp if everything went OK
- write the nagios check file
"""
options = {
'nagios-check-interval-threshold':
NAGIOS_CHECK_INTERVAL_THRESHOLD,
'storage': ('storage systems on which to deploy users and vos', None,
'extend', []),
'user': ('process users', None, 'store_true', False),
'vo': ('process vos', None, 'store_true', False),
'access_token': ('OAuth2 token to access the account page REST API',
None, 'store', None),
'account_page_url':
('URL of the account page where we can find the REST API', None,
'store', None),
'host_institute':
('Name of the institute where this script is being run', str, 'store',
GENT),
}
opts = ExtendedSimpleOption(options)
stats = {}
try:
now = datetime.utcnow()
client = AccountpageClient(token=opts.options.access_token,
url=opts.options.account_page_url + "/api/")
try:
last_timestamp = read_timestamp(SYNC_TIMESTAMP_FILENAME)
except Exception:
logger.exception("Something broke reading the timestamp from %s" %
SYNC_TIMESTAMP_FILENAME)
last_timestamp = "200901010000Z"
logger.info("Last recorded timestamp was %s" % (last_timestamp))
last_timestamp = convert_to_unix_timestamp(last_timestamp)
(users_ok, users_fail) = ([], [])
(quota_ok, quota_fail) = ([], [])
if opts.options.user:
ugent_changed_accounts = client.account.institute['gent'].modified[
last_timestamp].get()[1]
logger.info(
"Found %d UGent accounts that have changed in the accountpage since %s"
% (len(ugent_changed_accounts), last_timestamp))
ugent_accounts = [u['vsc_id'] for u in ugent_changed_accounts]
ugent_accounts = nub(ugent_accounts)
for storage_name in opts.options.storage:
(users_ok,
users_fail) = process_users(opts.options, ugent_accounts,
storage_name, client,
opts.options.host_institute)
stats["%s_users_sync" % (storage_name, )] = len(users_ok)
stats["%s_users_sync_fail" %
(storage_name, )] = len(users_fail)
stats["%s_users_sync_fail_warning" %
(storage_name, )] = STORAGE_USERS_LIMIT_WARNING
stats["%s_users_sync_fail_critical" %
(storage_name, )] = STORAGE_USERS_LIMIT_CRITICAL
for storage_name in opts.options.storage:
storage_changed_quota = [
mkVscUserSizeQuota(q) for q in client.quota.user.
storage[storage_name].modified[last_timestamp].get()[1]
]
storage_changed_quota = [
q for q in storage_changed_quota
if q.fileset.startswith('vsc')
]
logger.info(
"Found %d accounts that have changed quota on storage %s in the accountpage since %s",
len(storage_changed_quota), storage_name, last_timestamp)
(quota_ok, quota_fail) = process_users_quota(
opts.options, storage_changed_quota, storage_name, client,
opts.options.host_institute)
stats["%s_quota_sync" % (storage_name, )] = len(quota_ok)
stats["%s_quota_sync_fail" %
(storage_name, )] = len(quota_fail)
stats["%s_quota_sync_fail_warning" %
(storage_name, )] = STORAGE_QUOTA_LIMIT_WARNING
stats["%s_quota_sync_fail_critical" %
(storage_name, )] = STORAGE_QUOTA_LIMIT_CRITICAL
(vos_ok, vos_fail) = ([], [])
if opts.options.vo:
ugent_changed_vos = client.vo.modified[last_timestamp].get()[1]
ugent_changed_vo_quota = client.quota.vo.modified[
last_timestamp].get()[1]
ugent_vos = sorted(
set([v['vsc_id'] for v in ugent_changed_vos] + [
v['virtual_organisation'] for v in ugent_changed_vo_quota
]))
logger.info(
"Found %d UGent VOs that have changed in the accountpage since %s"
% (len(ugent_changed_vos), last_timestamp))
logger.info(
"Found %d UGent VOs that have changed quota in the accountpage since %s"
% (len(ugent_changed_vo_quota), last_timestamp))
logger.debug(
"Found the following UGent VOs: {vos}".format(vos=ugent_vos))
for storage_name in opts.options.storage:
(vos_ok, vos_fail) = process_vos(opts.options, ugent_vos,
storage_name, client,
last_timestamp,
opts.options.host_institute)
stats["%s_vos_sync" % (storage_name, )] = len(vos_ok)
stats["%s_vos_sync_fail" % (storage_name, )] = len(vos_fail)
stats["%s_vos_sync_fail_warning" %
(storage_name, )] = STORAGE_VO_LIMIT_WARNING
stats["%s_vos_sync_fail_critical" %
(storage_name, )] = STORAGE_VO_LIMIT_CRITICAL
if not (users_fail or quota_fail or vos_fail):
(_, ldap_timestamp) = convert_timestamp(now)
if not opts.options.dry_run:
write_timestamp(SYNC_TIMESTAMP_FILENAME, ldap_timestamp)
except Exception as err:
logger.exception("critical exception caught: %s" % (err))
opts.critical("Script failed in a horrible way")
sys.exit(NAGIOS_EXIT_CRITICAL)
opts.epilogue("UGent users and VOs synchronised", stats)
def main():
"""
Main script. The usual.
"""
options = {
"nagios-check-interval-threshold":
NAGIOS_CHECK_INTERVAL_THRESHOLD,
"access_token": ("OAuth2 token to access the account page REST API",
None, "store", None),
"account_page_url": (
"URL of the account page where we can find the REST API",
str,
"store",
"https://apivsc.ugent.be/django",
),
'host_institute':
('Name of the institute where this script is being run', str, 'store',
GENT),
"clusters": (
"Cluster(s) (comma-separated) to sync for. "
"Overrides <host_institute>_SLURM_COMPUTE_CLUSTERS that are in production.",
"strlist",
"store",
[],
),
'start_timestamp':
('Timestamp to start the sync from', str, 'store', None),
'cluster_classes':
('Classes of clusters that should be synced, comma-separated',
"strlist", 'store', [PRODUCTION, PILOT])
}
opts = ExtendedSimpleOption(options)
stats = {}
(last_timestamp, start_time) = retrieve_timestamp_with_default(
SYNC_TIMESTAMP_FILENAME, start_timestamp=opts.options.start_timestamp)
logging.info("Using timestamp %s", last_timestamp)
logging.info("Using startime %s", start_time)
try:
client = AccountpageClient(token=opts.options.access_token,
url=opts.options.account_page_url + "/api/")
host_institute = opts.options.host_institute
slurm_account_info = get_slurm_acct_info(SyncTypes.accounts)
slurm_user_info = get_slurm_acct_info(SyncTypes.users)
logging.debug("%d accounts found", len(slurm_account_info))
logging.debug("%d users found", len(slurm_user_info))
if opts.options.clusters:
clusters = opts.options.clusters
else:
clusters = [
cs for p in opts.options.cluster_classes
for cs in VSC_SLURM_CLUSTERS[host_institute][p]
]
sacctmgr_commands = []
# All users belong to a VO, so fetching the VOs is necessary/
account_page_vos = [
mkVo(v)
for v in client.vo.institute[opts.options.host_institute].get()[1]
]
# make sure the institutes and the default accounts (VOs) are there for each cluster
institute_vos = dict([
(v.vsc_id, v) for v in account_page_vos
if v.vsc_id in INSTITUTE_VOS_BY_INSTITUTE[host_institute].values()
])
sacctmgr_commands += slurm_institute_accounts(slurm_account_info,
clusters, host_institute,
institute_vos)
# The VOs do not track active state of users, so we need to fetch all accounts as well
active_accounts = set(
[a["vsc_id"] for a in client.account.get()[1] if a["isactive"]])
# dictionary mapping the VO vsc_id on a tuple with the VO members and the VO itself
account_page_members = dict([(vo.vsc_id, (set(vo.members), vo))
for vo in account_page_vos])
# process all regular VOs
sacctmgr_commands += slurm_vo_accounts(account_page_vos,
slurm_account_info, clusters,
host_institute)
# process VO members
sacctmgr_commands += slurm_user_accounts(account_page_members,
active_accounts,
slurm_user_info, clusters,
opts.options.dry_run)
logging.info("Executing %d commands", len(sacctmgr_commands))
if opts.options.dry_run:
print("Commands to be executed:\n")
print("\n".join([" ".join(c) for c in sacctmgr_commands]))
else:
execute_commands(sacctmgr_commands)
if not opts.options.dry_run:
(_, ldap_timestamp) = convert_timestamp(start_time)
write_timestamp(SYNC_TIMESTAMP_FILENAME, ldap_timestamp)
opts.epilogue("Accounts synced to slurm", stats)
else:
logging.info("Dry run done")
except Exception as err:
logging.exception("critical exception caught: %s", err)
opts.critical("Script failed in a horrible way")
sys.exit(NAGIOS_EXIT_CRITICAL)
def main():
options = {
'nagios-check-interval-threshold':
NAGIOS_CHECK_INTERVAL_THRESHOLD,
'start-timestamp':
("The timestamp form which to start, otherwise use the cached value",
None, "store", None),
'access_token':
('OAuth2 token identifying the user with the accountpage', None,
'store', None),
'account_page_url': ('url for the account page', None, 'store', None),
'start_timestamp':
('Timestamp to start the sync from', str, 'store', None),
}
# get access_token from conf file
ExtendedSimpleOption.CONFIGFILES_INIT = ['/etc/account_page.conf']
opts = ExtendedSimpleOption(options)
stats = {}
# Creating this here because this is a singleton class
_ = LdapQuery(VscConfiguration(VSC_CONF_DEFAULT_FILENAME))
(last_timestamp, start_time) = retrieve_timestamp_with_default(
SYNC_TIMESTAMP_FILENAME, start_timestamp=opts.options.start_timestamp)
logging.info("Using timestamp %s", last_timestamp)
logging.info("Using startime %s", start_time)
try:
parent_pid = os.fork()
logging.info("Forked.")
except OSError:
logging.exception("Could not fork")
parent_pid = 1
except Exception:
logging.exception("Oops")
parent_pid = 1
if parent_pid == 0:
try:
global logger
logger = fancylogger.getLogger(NAGIOS_HEADER)
# drop privileges in the child
try:
apache_uid = pwd.getpwnam('apache').pw_uid
apache_gid = grp.getgrnam('apache').gr_gid
os.setgroups([])
os.setgid(apache_gid)
os.setuid(apache_uid)
logging.info("Now running as %s" % (os.geteuid(), ))
except OSError:
logger.raiseException("Could not drop privileges")
client = AccountpageClient(token=opts.options.access_token,
url=opts.options.account_page_url +
'/api/')
syncer = LdapSyncer(client)
last = last_timestamp
altered_accounts = syncer.sync_altered_accounts(
last, opts.options.dry_run)
logging.debug("Altered accounts: %s", altered_accounts)
altered_groups = syncer.sync_altered_groups(
last, opts.options.dry_run)
logging.debug("Altered groups: %s" % altered_groups)
if not altered_accounts[ERROR] \
and not altered_groups[ERROR]:
logging.info("Child process exiting correctly")
sys.exit(0)
else:
logging.info("Child process exiting with status -1")
logging.warning("Error occured in %s" % ([
"%s: %s\n" % (k, v) for (k, v) in [
("altered accounts", altered_accounts[ERROR]),
("altered groups", altered_groups[ERROR]),
]
]))
sys.exit(-1)
except Exception:
logging.exception("Child caught an exception")
sys.exit(-1)
else:
# parent
(_, result) = os.waitpid(parent_pid, 0)
logging.info("Child exited with exit code %d" % (result, ))
if not result and not opts.options.dry_run:
(_, ldap_timestamp) = convert_timestamp(start_time)
write_timestamp(SYNC_TIMESTAMP_FILENAME, ldap_timestamp)
opts.epilogue("Synchronised LDAP users to the Django DB", stats)
else:
sys.exit(NAGIOS_EXIT_CRITICAL)
def main():
options = {
'nagios-check-interval-threshold':
NAGIOS_CHECK_INTERVAL_THRESHOLD,
'start-timestamp':
("The timestamp form which to start, otherwise use the cached value",
None, "store", None),
'access_token':
('OAuth2 token identifying the user with the accountpage', None,
'store', None),
'account_page_url': ('url for the account page', None, 'store', None),
}
# get access_token from conf file
ExtendedSimpleOption.CONFIGFILES_INIT = ['/etc/account_page.conf']
opts = ExtendedSimpleOption(options)
stats = {}
# Creating this here because this is a singleton class
_ = LdapQuery(VscConfiguration(VSC_CONF_DEFAULT_FILENAME))
last_timestamp = opts.options.start_timestamp
if not last_timestamp:
try:
last_timestamp = read_timestamp(SYNC_TIMESTAMP_FILENAME)
except Exception:
_log.warning("Something broke reading the timestamp from %s",
SYNC_TIMESTAMP_FILENAME)
last_timestamp = "201710230000Z"
_log.warning(
"We will resync from a hardcoded know working sync a while back : %s",
last_timestamp)
_log.info("Using timestamp %s", last_timestamp)
# record starttime before starting, and take a 10 sec safety buffer so we don't get gaps where users are approved
# in between the requesting of modified users and writing out the start time
start_time = datetime.datetime.now() + datetime.timedelta(seconds=-10)
_log.info("startime %s", start_time)
try:
parent_pid = os.fork()
_log.info("Forked.")
except OSError:
_log.exception("Could not fork")
parent_pid = 1
except Exception:
_log.exception("Oops")
parent_pid = 1
if parent_pid == 0:
try:
global _log
_log = fancylogger.getLogger(NAGIOS_HEADER)
# drop privileges in the child
try:
apache_uid = pwd.getpwnam('apache').pw_uid
apache_gid = grp.getgrnam('apache').gr_gid
os.setgroups([])
os.setgid(apache_gid)
os.setuid(apache_uid)
_log.info("Now running as %s" % (os.geteuid(), ))
except OSError:
_log.raiseException("Could not drop privileges")
client = AccountpageClient(token=opts.options.access_token,
url=opts.options.account_page_url +
'/api/')
syncer = LdapSyncer(client)
last = int(
(datetime.datetime.strptime(last_timestamp, "%Y%m%d%H%M%SZ") -
datetime.datetime(1970, 1, 1)).total_seconds())
altered_accounts = syncer.sync_altered_accounts(
last, opts.options.dry_run)
_log.debug("Altered accounts: %s", altered_accounts)
altered_groups = syncer.sync_altered_groups(
last, opts.options.dry_run)
_log.debug("Altered groups: %s" % altered_groups)
if not altered_accounts[ERROR] \
and not altered_groups[ERROR]:
_log.info("Child process exiting correctly")
sys.exit(0)
else:
_log.info("Child process exiting with status -1")
_log.warning("Error occured in %s" % ([
"%s: %s\n" % (k, v) for (k, v) in [
("altered accounts", altered_accounts[ERROR]),
("altered groups", altered_groups[ERROR]),
]
]))
sys.exit(-1)
except Exception:
_log.exception("Child caught an exception")
sys.exit(-1)
else:
# parent
(_, result) = os.waitpid(parent_pid, 0)
_log.info("Child exited with exit code %d" % (result, ))
if not result:
if not opts.options.start_timestamp:
(_, ldap_timestamp) = convert_timestamp(start_time)
if not opts.options.dry_run:
write_timestamp(SYNC_TIMESTAMP_FILENAME, ldap_timestamp)
else:
_log.info(
"Not updating the timestamp, since one was provided on the command line"
)
opts.epilogue("Synchronised LDAP users to the Django DB", stats)
else:
_log.info(
"Not updating the timestamp, since it was given on the command line for this run"
)
sys.exit(NAGIOS_EXIT_CRITICAL)
def main():
"""
Main script.
- build the filter
- fetches the users
- process the users
- write the new timestamp if everything went OK
- write the nagios check file
"""
options = {
'nagios-check-interval-threshold':
NAGIOS_CHECK_INTERVAL_THRESHOLD,
'storage': ('storage systems on which to deploy users and vos', None,
'extend', []),
'user': ('process users', None, 'store_true', False),
'vo': ('process vos', None, 'store_true', False),
'access_token': ('OAuth2 token to access the account page REST API',
None, 'store', None),
'account_page_url':
('URL of the account page where we can find the REST API', None,
'store', None),
'host_institute':
('Name of the institute where this script is being run', str, 'store',
GENT),
'start_timestamp':
('Timestamp to start the sync from', str, 'store', None),
}
opts = ExtendedSimpleOption(options)
stats = {}
(last_timestamp, start_time) = retrieve_timestamp_with_default(
SYNC_TIMESTAMP_FILENAME, start_timestamp=opts.options.start_timestamp)
logging.info("Using timestamp %s", last_timestamp)
logging.info("Using startime %s", start_time)
try:
client = AccountpageClient(token=opts.options.access_token,
url=opts.options.account_page_url + "/api/")
institute = opts.options.host_institute
(users_ok, users_fail) = ([], [])
(quota_ok, quota_fail) = ([], [])
if opts.options.user:
changed_accounts = client.account.institute[institute].modified[
last_timestamp].get()[1]
logging.info(
"Found %d %s accounts that have changed in the accountpage since %s"
% (len(changed_accounts), institute, last_timestamp))
accounts = nub([u['vsc_id'] for u in changed_accounts])
for storage_name in opts.options.storage:
(users_ok, users_fail) = process_users(opts.options, accounts,
storage_name, client,
institute)
stats["%s_users_sync" % (storage_name, )] = len(users_ok)
stats["%s_users_sync_fail" %
(storage_name, )] = len(users_fail)
stats["%s_users_sync_fail_warning" %
(storage_name, )] = STORAGE_USERS_LIMIT_WARNING
stats["%s_users_sync_fail_critical" %
(storage_name, )] = STORAGE_USERS_LIMIT_CRITICAL
for storage_name in opts.options.storage:
storage_changed_quota = [
mkVscUserSizeQuota(q) for q in client.quota.user.
storage[storage_name].modified[last_timestamp].get()[1]
]
storage_changed_quota = [
q for q in storage_changed_quota
if q.fileset.startswith('vsc')
]
logging.info(
"Found %d accounts that have changed quota on storage %s in the accountpage since %s",
len(storage_changed_quota), storage_name, last_timestamp)
(quota_ok, quota_fail) = process_users_quota(
opts.options, storage_changed_quota, storage_name, client,
institute)
stats["%s_quota_sync" % (storage_name, )] = len(quota_ok)
stats["%s_quota_sync_fail" %
(storage_name, )] = len(quota_fail)
stats["%s_quota_sync_fail_warning" %
(storage_name, )] = STORAGE_QUOTA_LIMIT_WARNING
stats["%s_quota_sync_fail_critical" %
(storage_name, )] = STORAGE_QUOTA_LIMIT_CRITICAL
(vos_ok, vos_fail) = ([], [])
if opts.options.vo:
changed_vos = client.vo.institute[institute].modified[
last_timestamp].get()[1]
changed_vo_quota = client.quota.vo.modified[last_timestamp].get(
)[1]
vos = sorted(
set([v['vsc_id'] for v in changed_vos] +
[v['virtual_organisation'] for v in changed_vo_quota]))
logging.info(
"Found %d %s VOs that have changed in the accountpage since %s"
% (len(changed_vos), institute, last_timestamp))
logging.info(
"Found %d %s VOs that have changed quota in the accountpage since %s"
% (len(changed_vo_quota), institute, last_timestamp))
logging.debug("Found the following {institute} VOs: {vos}".format(
institute=institute, vos=vos))
for storage_name in opts.options.storage:
(vos_ok, vos_fail) = process_vos(opts.options, vos,
storage_name, client,
last_timestamp, institute)
stats["%s_vos_sync" % (storage_name, )] = len(vos_ok)
stats["%s_vos_sync_fail" % (storage_name, )] = len(vos_fail)
stats["%s_vos_sync_fail_warning" %
(storage_name, )] = STORAGE_VO_LIMIT_WARNING
stats["%s_vos_sync_fail_critical" %
(storage_name, )] = STORAGE_VO_LIMIT_CRITICAL
if not (users_fail or quota_fail
or vos_fail) and not opts.options.dry_run:
(_, ldap_timestamp) = convert_timestamp(start_time)
write_timestamp(SYNC_TIMESTAMP_FILENAME, ldap_timestamp)
except Exception as err:
logger.exception("critical exception caught: %s" % (err))
opts.critical("Script failed in a horrible way")
sys.exit(NAGIOS_EXIT_CRITICAL)
opts.epilogue("%s users and VOs synchronised" % institute, stats)
def main():
"""Main script"""
options = {
'nagios-check-interval-threshold':
NAGIOS_CHECK_INTERVAL_THRESHOLD,
'storage': ('the VSC filesystems that are checked by this script',
None, 'extend', []),
'write-cache': ('Write the data into the cache files in the FS', None,
'store_true', False),
'account_page_url': ('Base URL of the account page', None, 'store',
'https://account.vscentrum.be/django'),
'access_token': ('OAuth2 token to access the account page REST API',
None, 'store', None),
'host_institute':
('Name of the institute where this script is being run', str, 'store',
GENT),
}
opts = ExtendedSimpleOption(options)
logger = opts.log
try:
client = AccountpageClient(token=opts.options.access_token)
user_id_map = map_uids_to_names() # is this really necessary?
gpfs = GpfsOperations()
storage = VscStorage()
target_filesystems = [
storage[s].filesystem for s in opts.options.storage
]
filesystems = gpfs.list_filesystems(device=target_filesystems).keys()
logger.debug("Found the following GPFS filesystems: %s" %
(filesystems))
filesets = gpfs.list_filesets(devices=target_filesystems)
logger.debug("Found the following GPFS filesets: %s" % (filesets))
quota = gpfs.list_quota(devices=target_filesystems)
exceeding_filesets = {}
exceeding_users = {}
stats = {}
for storage_name in opts.options.storage:
logger.info("Processing quota for storage_name %s" %
(storage_name))
filesystem = storage[storage_name].filesystem
replication_factor = storage[storage_name].data_replication_factor
if filesystem not in filesystems:
logger.error("Non-existent filesystem %s" % (filesystem))
continue
if filesystem not in quota.keys():
logger.error("No quota defined for storage_name %s [%s]" %
(storage_name, filesystem))
continue
quota_storage_map = get_mmrepquota_maps(
quota[filesystem],
storage_name,
filesystem,
filesets,
replication_factor,
)
exceeding_filesets[storage_name] = process_fileset_quota(
storage,
gpfs,
storage_name,
filesystem,
quota_storage_map['FILESET'],
client,
dry_run=opts.options.dry_run,
institute=opts.options.host_institute)
exceeding_users[storage_name] = process_user_quota(
storage,
gpfs,
storage_name,
None,
quota_storage_map['USR'],
user_id_map,
client,
dry_run=opts.options.dry_run,
institute=opts.options.host_institute)
stats["%s_fileset_critical" %
(storage_name, )] = QUOTA_FILESETS_CRITICAL
if exceeding_filesets[storage_name]:
stats["%s_fileset" % (storage_name, )] = 1
logger.warning(
"storage_name %s found %d filesets that are exceeding their quota",
storage_name, len(exceeding_filesets))
for (e_fileset, e_quota) in exceeding_filesets[storage_name]:
logger.warning("%s has quota %s" %
(e_fileset, str(e_quota)))
else:
stats["%s_fileset" % (storage_name, )] = 0
logger.debug(
"storage_name %s found no filesets that are exceeding their quota"
% storage_name)
stats["%s_users_warning" % (storage_name, )] = QUOTA_USERS_WARNING
stats["%s_users_critical" %
(storage_name, )] = QUOTA_USERS_CRITICAL
if exceeding_users[storage_name]:
stats["%s_users" % (storage_name, )] = len(
exceeding_users[storage_name])
logger.warning(
"storage_name %s found %d users who are exceeding their quota"
% (storage_name, len(exceeding_users[storage_name])))
for (e_user_id, e_quota) in exceeding_users[storage_name]:
logger.warning("%s has quota %s" %
(e_user_id, str(e_quota)))
else:
stats["%s_users" % (storage_name, )] = 0
logger.debug(
"storage_name %s found no users who are exceeding their quota"
% storage_name)
except Exception as err:
logger.exception("critical exception caught: %s" % (err))
opts.critical("Script failed in a horrible way")
opts.epilogue("quota check completed", stats)
def main():
# Collect all info
# Note: debug option is provided by generaloption
# Note: other settings, e.g., ofr each cluster will be obtained from the configuration file
options = {
'nagios-check-interval-threshold':
NAGIOS_CHECK_INTERVAL_THRESHOLD,
'hosts':
('the hosts/clusters that should be contacted for job information',
None, 'extend', []),
'location': ('the location for storing the pickle file: delcatty, muk',
str, 'store', 'delcatty'),
'access_token':
('the token that will allow authentication against the account page',
None, 'store', None),
'account_page_url': ('', None, 'store', None),
'target_master':
('the master used to execute showq commands', None, 'store', None),
'target_user':
('the user for ssh to the target master', None, 'store', None),
}
opts = ExtendedSimpleOption(options)
try:
rest_client = AccountpageClient(token=opts.options.access_token)
gpfs = GpfsOperations()
storage = VscStorage()
storage_name = cluster_user_pickle_store_map[opts.options.location]
login_mount_point = storage[storage_name].login_mount_point
gpfs_mount_point = storage[storage_name].gpfs_mount_point
clusters = {}
for host in opts.options.hosts:
master = opts.configfile_parser.get(host, "master")
checkjob_path = opts.configfile_parser.get(host, "checkjob_path")
clusters[host] = {'master': master, 'path': checkjob_path}
checkjob = SshCheckjob(opts.options.target_master,
opts.options.target_user,
clusters,
cache_pickle=True,
dry_run=opts.options.dry_run)
(job_information, _, _) = checkjob.get_moab_command_information()
active_users = job_information.keys()
logger.debug("Active users: %s" % (active_users))
logger.debug("Checkjob information: %s" % (job_information))
nagios_user_count = 0
nagios_no_store = 0
stats = {}
for user in active_users:
path = get_pickle_path(opts.options.location, user, rest_client)
try:
user_queue_information = CheckjobInfo(
{user: job_information[user]})
store_on_gpfs(user, path, "checkjob", user_queue_information,
gpfs, login_mount_point, gpfs_mount_point,
".checkjob.json.gz", opts.options.dry_run)
nagios_user_count += 1
except Exception:
logger.exception("Could not store cache file for user %s" %
(user))
nagios_no_store += 1
stats["store_users"] = nagios_user_count
stats["store_fail"] = nagios_no_store
stats["store_fail_critical"] = STORE_LIMIT_CRITICAL
except Exception, err:
logger.exception("critical exception caught: %s" % (err))
opts.critical("Script failed in a horrible way")
sys.exit(NAGIOS_EXIT_CRITICAL)
def main():
"""
Main script. The usual.
"""
options = {
"nagios-check-interval-threshold":
NAGIOS_CHECK_INTERVAL_THRESHOLD,
"access_token": ("OAuth2 token to access the account page REST API",
None, "store", None),
"account_page_url": (
"URL of the account page where we can find the REST API",
str,
"store",
"https://apivsc.ugent.be/django",
),
"clusters": (
"Cluster(s) (comma-separated) to sync for. "
"Overrides GENT_SLURM_COMPUTE_CLUSTERS that are in production.",
str,
"store",
None,
),
}
opts = ExtendedSimpleOption(options)
stats = {}
try:
client = AccountpageClient(token=opts.options.access_token,
url=opts.options.account_page_url + "/api/")
last_timestamp = "201804010000Z" # the beginning of time
logging.info("Last recorded timestamp was %s" % (last_timestamp))
slurm_account_info = get_slurm_acct_info(SyncTypes.accounts)
slurm_user_info = get_slurm_acct_info(SyncTypes.users)
logging.debug("%d accounts found", len(slurm_account_info))
logging.debug("%d users found", len(slurm_user_info))
if opts.options.clusters is not None:
clusters = opts.options.clusters.split(",")
else:
clusters = [
c for c in GENT_SLURM_COMPUTE_CLUSTERS
if c in GENT_PRODUCTION_COMPUTE_CLUSTERS
]
sacctmgr_commands = []
# make sure the institutes and the default accounts (VOs) are there for each cluster
sacctmgr_commands += slurm_institute_accounts(slurm_account_info,
clusters)
# All users belong to a VO, so fetching the VOs is necessary/
account_page_vos = [mkVo(v) for v in client.vo.get()[1]]
# The VOs do not track active state of users, so we need to fetch all accounts as well
active_accounts = set(
[a["vsc_id"] for a in client.account.get()[1] if a["isactive"]])
# dictionary mapping the VO vsc_id on a tuple with the VO members and the VO itself
account_page_members = dict([(vo.vsc_id, (set(vo.members), vo))
for vo in account_page_vos])
# process all regular VOs
sacctmgr_commands += slurm_vo_accounts(account_page_vos,
slurm_account_info, clusters)
# process VO members
sacctmgr_commands += slurm_user_accounts(account_page_members,
active_accounts,
slurm_user_info, clusters,
opts.options.dry_run)
logging.info("Executing %d commands", len(sacctmgr_commands))
if opts.options.dry_run:
print("Commands to be executed:\n")
print("\n".join([" ".join(c) for c in sacctmgr_commands]))
else:
execute_commands(sacctmgr_commands)
except Exception as err:
logger.exception("critical exception caught: %s" % (err))
opts.critical("Script failed in a horrible way")
sys.exit(NAGIOS_EXIT_CRITICAL)
if not opts.options.dry_run:
opts.epilogue("Accounts synced to slurm", stats)
else:
logger.info("Dry run done")
def main():
# Collect all info
# Note: debug option is provided by generaloption
# Note: other settings, e.g., ofr each cluster will be obtained from the configuration file
options = {
'nagios-check-interval-threshold':
NAGIOS_CHECK_INTERVAL_THRESHOLD,
'hosts':
('the hosts/clusters that should be contacted for job information',
None, 'extend', []),
'information': ('the sort of information to store: user, vo, project',
None, 'store', 'user'),
'location': ('the location for storing the pickle file: delcatty, muk',
str, 'store', 'delcatty'),
'account_page_url':
('the URL at which the account page resides', None, 'store', None),
'access_token':
('the token that will allow authentication against the account page',
None, 'store', None),
'target_master':
('the master used to execute showq commands', None, 'store', None),
'target_user': ('the user for ssh to the target master', None, 'store',
None),
}
opts = ExtendedSimpleOption(options)
try:
rest_client = AccountpageClient(token=opts.options.access_token)
gpfs = GpfsOperations()
storage = VscStorage()
storage_name = cluster_user_pickle_store_map[opts.options.location]
login_mount_point = storage[storage_name].login_mount_point
gpfs_mount_point = storage[storage_name].gpfs_mount_point
clusters = {}
for host in opts.options.hosts:
master = opts.configfile_parser.get(host, "master")
showq_path = opts.configfile_parser.get(host, "showq_path")
clusters[host] = {'master': master, 'path': showq_path}
logger.debug("clusters = %s" % (clusters, ))
showq = SshShowq(opts.options.target_master,
opts.options.target_user,
clusters,
cache_pickle=True,
dry_run=opts.options.dry_run)
logger.debug("Getting showq information ...")
(queue_information, _, _) = showq.get_moab_command_information()
timeinfo = time.time()
active_users = queue_information.keys()
logger.debug("Active users: %s" % (active_users))
logger.debug("Queue information: %s" % (queue_information))
# We need to determine which users should get an updated pickle. This depends on
# - the active user set
# - the information we want to provide on the cluster(set) where this script runs
# At the same time, we need to determine the job information each user gets to see
tup = (opts.options.information, active_users, queue_information,
rest_client)
(target_users, target_queue_information,
user_map) = determine_target_information(*tup)
nagios_user_count = 0
nagios_no_store = 0
stats = {}
for user in target_users:
try:
path = get_pickle_path(opts.options.location, user,
rest_client)
user_queue_information = target_queue_information[user]
user_queue_information['timeinfo'] = timeinfo
store_on_gpfs(user, path, "showq",
(user_queue_information, user_map[user]), gpfs,
login_mount_point, gpfs_mount_point,
".showq.json.gz", opts.options.dry_run)
nagios_user_count += 1
except Exception:
logger.error("Could not store pickle file for user %s" %
(user))
nagios_no_store += 1
stats["store_users"] = nagios_user_count
stats["store_fail"] = nagios_no_store
stats["store_fail_critical"] = STORE_LIMIT_CRITICAL
except Exception, err:
logger.exception("critical exception caught: %s" % (err))
opts.critical("Script failed in a horrible way")
sys.exit(NAGIOS_EXIT_CRITICAL)
def get_vsc_record(username, vsc_token, logger=None):
"""
Retrieve and update list of VSC users with data from VSC account page
- username: (string) VSC ID or institute user of the VSC account
- vsc_token: (string) access token to VSC account page
- logger: (object) fancylogger object of the caller
"""
if logger is None:
logger = fancylogger.getLogger()
vsc_api_client = AccountpageClient(token=vsc_token)
# Get institute login of the VSC account attached to this username
if username[0:3] == 'vsc' and username[3].isdigit():
# VSC ID: query institute login to VSC account page
logger.debug(f"[{username}] user treated as VSC ID")
try:
vsc_account = vsc_api_client.account[username].person.get()[1]
except HTTPError as err:
if err.code == 404:
error_exit(
logger,
f"[{username}] VSC ID not found in VSC account page")
else:
error_exit(logger, f"[{username}] {err}")
except (TimeoutError, URLError) as err:
error_exit(
logger,
f"[{username}] connection to VSC account page timed out")
else:
vsc_login = {
'username': vsc_account['institute_login'],
'site': vsc_account['institute']['name']
}
logger.debug(
f"[{username}] VSC ID belongs to VSC account '{vsc_login['username']}'"
)
else:
# Others: assume NetID from Brussels
logger.debug(f"[{username}] user treated as NetID")
vsc_login = {'username': username, 'site': BRUSSEL}
# Retrieve user data from VSC account page
try:
vsc_account = vsc_api_client.account.institute[vsc_login['site']].id[
vsc_login['username']].get()[1]
except HTTPError as err:
if err.code == 404:
logger.debug(
f"[{username}] with VSC account '{vsc_login['username']}' not found"
)
return None
else:
error_exit(logger, f"[{username}] {err}")
except (TimeoutError, URLError) as err:
error_exit(logger,
f"[{username}] connection to VSC account page timed out")
else:
logger.debug(
f"[{username}] user account record retrieved from VSC account '{vsc_login['username']}'"
)
user_record = {
# only use first entry of research field
'field': vsc_account['research_field'][0],
'site':
INSTITUTE_LONGNAME[vsc_account['person']['institute']['name']],
'updated': date.today().isoformat(),
}
return user_record