def __update_bulk_replicas(replicas, session=None, logger=logging.log):
"""
Used by finisher to handle available and unavailable replicas blongs to same rule in bulk way.
:param replicas: List of replicas.
:param session: The database session to use.
:returns commit_or_rollback: Boolean.
"""
try:
replica_core.update_replicas_states(replicas,
nowait=True,
add_tombstone=True,
session=session)
except ReplicaNotFound as error:
logger(logging.WARNING,
'Failed to bulk update replicas, will do it one by one: %s',
str(error))
raise ReplicaNotFound(error)
for replica in replicas:
if not replica['archived']:
request_core.archive_request(replica['request_id'],
session=session)
logger(logging.INFO, "HANDLED REQUEST %s DID %s:%s AT RSE %s STATE %s",
replica['request_id'], replica['scope'], replica['name'],
replica['rse_id'], str(replica['state']))
return True
def __update_bulk_replicas(replicas, req_type, rule_id, session=None):
"""
Used by finisher to handle available and unavailable replicas blongs to same rule in bulk way.
:param replicas: List of replicas.
:param req_type: Request type: STAGEIN, STAGEOUT, TRANSFER.
:param rule_id: RULE id.
:param session: The database session to use.
:returns commit_or_rollback: Boolean.
"""
try:
replica_core.update_replicas_states(replicas,
nowait=True,
session=session)
except ReplicaNotFound as error:
logging.warn(
'Failed to bulk update replicas, will do it one by one: %s' %
str(error))
raise ReplicaNotFound(error)
for replica in replicas:
if not replica['archived']:
request_core.archive_request(replica['request_id'],
session=session)
logging.info("HANDLED REQUEST %s DID %s:%s AT RSE %s STATE %s" %
(replica['request_id'], replica['scope'], replica['name'],
replica['rse_id'], str(replica['state'])))
return True
def GET(self, scope, name):
"""
Metalink redirect
HTTP Success:
200 OK
HTTP Error:
401 Unauthorized
500 InternalError
404 Notfound
406 Not Acceptable
:param scope: The scope name of the file.
:param name: The name of the file.
"""
header('Access-Control-Allow-Origin', ctx.env.get('HTTP_ORIGIN'))
header('Access-Control-Allow-Headers',
ctx.env.get('HTTP_ACCESS_CONTROL_REQUEST_HEADERS'))
header('Access-Control-Allow-Methods', '*')
header('Access-Control-Allow-Credentials', 'true')
dids, schemes, select = [{
'scope': scope,
'name': name
}], [
'http', 'https', 's3+rucio', 's3+https', 'root', 'gsiftp', 'srm',
'davs'
], None
# set the correct client IP
client_ip = ctx.env.get('HTTP_X_FORWARDED_FOR')
if client_ip is None:
client_ip = ctx.ip
client_location = {'ip': client_ip, 'fqdn': None, 'site': None}
if ctx.query:
params = parse_qs(ctx.query[1:])
if 'schemes' in params:
schemes = params['schemes']
if 'select' in params:
select = params['select'][0]
if 'sort' in params:
select = params['sort'][0]
if 'ip' in params:
client_location['ip'] = params['ip'][0]
if 'fqdn' in params:
client_location['fqdn'] = params['fqdn'][0]
if 'site' in params:
client_location['site'] = params['site'][0]
try:
tmp_replicas = [
rep for rep in list_replicas(dids=dids,
schemes=schemes,
client_location=client_location)
]
if not tmp_replicas:
raise ReplicaNotFound(
'no redirection possible - cannot find the DID')
# first, set the appropriate content type, and stream the header
header('Content-Type', 'application/metalink4+xml')
yield '<?xml version="1.0" encoding="UTF-8"?>\n<metalink xmlns="urn:ietf:params:xml:ns:metalink">\n'
# iteratively stream the XML per file
for rfile in tmp_replicas:
replicas = []
dictreplica = {}
for rse in rfile['rses']:
for replica in rfile['rses'][rse]:
replicas.append(replica)
dictreplica[replica] = rse
# stream metadata
yield ' <file name="' + rfile['name'] + '">\n'
yield ' <identity>' + rfile['scope'] + ':' + rfile[
'name'] + '</identity>\n'
if rfile['adler32'] is not None:
yield ' <hash type="adler32">' + rfile[
'adler32'] + '</hash>\n'
if rfile['md5'] is not None:
yield ' <hash type="md5">' + rfile['md5'] + '</hash>\n'
yield ' <size>' + str(rfile['bytes']) + '</size>\n'
yield ' <glfn name="/atlas/rucio/%s:%s">' % (rfile['scope'],
rfile['name'])
yield '</glfn>\n'
# sort the actual replicas if necessary
if select == 'geoip':
replicas = sort_geoip(dictreplica,
client_location['ip'],
ignore_error=True)
elif select == 'closeness':
replicas = sort_closeness(dictreplica, client_location)
elif select == 'dynamic':
replicas = sort_dynamic(dictreplica, client_location)
elif select == 'ranking':
replicas = sort_ranking(dictreplica, client_location)
else:
replicas = sort_random(dictreplica)
# stream URLs
idx = 1
for replica in replicas:
yield ' <url location="' + str(
dictreplica[replica]) + '" priority="' + str(
idx) + '">' + replica + '</url>\n'
idx += 1
yield ' </file>\n'
# don't forget to send the metalink footer
yield '</metalink>\n'
except DataIdentifierNotFound as error:
raise generate_http_error(404, 'DataIdentifierNotFound',
error.args[0])
except ReplicaNotFound as error:
raise generate_http_error(404, 'ReplicaNotFound', error.args[0])
except RucioException as error:
raise generate_http_error(500, error.__class__.__name__,
error.args[0])
except Exception as error:
print(format_exc())
raise InternalError(error)
def GET(self, scope, name):
"""
Header Redirect
HTTP Success:
303 See Other
HTTP Error:
401 Unauthorized
500 InternalError
404 Notfound
:param scope: The scope name of the file.
:param name: The name of the file.
"""
header('Access-Control-Allow-Origin', ctx.env.get('HTTP_ORIGIN'))
header('Access-Control-Allow-Headers',
ctx.env.get('HTTP_ACCESS_CONTROL_REQUEST_HEADERS'))
header('Access-Control-Allow-Methods', '*')
header('Access-Control-Allow-Credentials', 'true')
try:
# use the default HTTP protocols if no scheme is given
select, rse, site, schemes = 'random', None, None, [
'davs', 'http', 'https', 's3+rucio'
]
client_ip = ctx.env.get('HTTP_X_FORWARDED_FOR')
if client_ip is None:
client_ip = ctx.ip
client_location = {'ip': client_ip, 'fqdn': None, 'site': None}
if ctx.query:
params = parse_qs(ctx.query[1:])
if 'select' in params:
select = params['select'][0]
if 'sort' in params:
select = params['sort'][0]
if 'rse' in params:
rse = params['rse'][0]
if 'site' in params:
site = params['site'][0]
if 'schemes' in params:
schemes = params['schemes'][0]
else:
schemes = ['davs', 'https', 's3']
if 'ip' in params:
client_location['ip'] = params['ip'][0]
if 'fqdn' in params:
client_location['fqdn'] = params['fqdn'][0]
if 'site' in params:
client_location['site'] = params['site'][0]
# correctly forward the schemes and select to potential metalink followups
cleaned_url = ctx.env.get('REQUEST_URI').split('?')[0]
if isinstance(schemes, list):
header(
'Link',
'<%s/metalink?schemes=%s&select=%s>; rel=describedby; type="application/metalink+xml"'
% (cleaned_url, ','.join(schemes), select))
else:
header(
'Link',
'<%s/metalink?schemes=%s&select=%s>; rel=describedby; type="application/metalink+xml"'
% (cleaned_url, schemes, select))
schemes = [schemes] # list_replicas needs a list
replicas = [
r for r in list_replicas(dids=[{
'scope': scope,
'name': name,
'type': 'FILE'
}],
schemes=schemes,
client_location=client_location)
]
selected_url, selected_rse = None, None
for r in replicas:
if r['rses']:
dictreplica = {}
if rse:
if rse in r['rses'] and r['rses'][rse]:
selected_url = r['rses'][rse][0]
selected_rse = rse
else:
raise ReplicaNotFound(
'no redirection possible - no valid RSE for HTTP redirection found'
)
else:
for rep in r['rses']:
for replica in r['rses'][rep]:
# since this is HTTP-only redirection, and to ensure compatibility with as many http clients as possible
# forcibly replacement davs and s3 URLs to https
replica = replica.replace(
'davs://',
'https://').replace('s3://', 'https://')
dictreplica[replica] = rep
if not dictreplica:
raise ReplicaNotFound(
'no redirection possible - no valid RSE for HTTP redirection found'
)
elif site:
rep = site_selector(dictreplica, site)
if rep:
selected_url = rep[0]
else:
raise ReplicaNotFound(
'no redirection possible - no valid RSE for HTTP redirection found'
)
else:
if select == 'geoip':
rep = sort_geoip(dictreplica,
client_location['ip'])
elif select == 'closeness':
rep = sort_closeness(dictreplica,
client_location)
elif select == 'dynamic':
rep = sort_dynamic(dictreplica,
client_location)
elif select == 'ranking':
rep = sort_ranking(dictreplica,
client_location)
else:
rep = sort_random(dictreplica)
selected_url = rep[0]
for rep in r['rses']:
for replica in r['rses'][rep]:
if selected_url == replica:
selected_rse = rep
if selected_url:
if selected_url.startswith('s3+rucio://'):
connect(selected_rse, selected_url)
signed_URLS = get_signed_urls([selected_url],
rse=selected_rse,
operation='read')
raise seeother(signed_URLS[selected_url])
raise seeother(selected_url)
raise ReplicaNotFound(
'no redirection possible - file does not exist')
except seeother:
raise
except ReplicaNotFound as error:
raise generate_http_error(404, 'ReplicaNotFound', error.args[0])
except RucioException as error:
raise generate_http_error(500, error.__class__.__name__,
error.args[0])
except Exception as error:
print(format_exc())
raise InternalError(error)
def GET(self, scope, name):
"""
Metalink redirect
HTTP Success:
200 OK
HTTP Error:
401 Unauthorized
500 InternalError
404 Notfound
:param scope: The scope name of the file.
:param name: The name of the file.
"""
header('Access-Control-Allow-Origin', ctx.env.get('HTTP_ORIGIN'))
header('Access-Control-Allow-Headers',
ctx.env.get('HTTP_ACCESS_CONTROL_REQUEST_HEADERS'))
header('Access-Control-Allow-Methods', '*')
header('Access-Control-Allow-Credentials', 'true')
dids, schemes, select = [{
'scope': scope,
'name': name
}], ['http', 'https', 's3+rucio', 's3+https', 'root', 'gsiftp',
'srm'], None
if ctx.query:
params = parse_qs(ctx.query[1:])
if 'schemes' in params:
schemes = params['schemes']
if 'select' in params:
select = params['select'][0]
try:
tmp_replicas = [
rep for rep in list_replicas(dids=dids, schemes=schemes)
]
if not tmp_replicas:
raise ReplicaNotFound(
'no redirection possible - cannot find the DID')
# first, set the APPropriate content type, and stream the header
header('Content-Type', 'application/metalink4+xml')
yield '<?xml version="1.0" encoding="UTF-8"?>\n<metalink xmlns="urn:ietf:params:xml:ns:metalink">\n'
# set the correct client IP
client_ip = ctx.env.get('HTTP_X_FORWARDED_FOR')
if client_ip is None:
client_ip = ctx.ip
# iteratively stream the XML per file
for rfile in tmp_replicas:
replicas = []
dictreplica = {}
for rse in rfile['rses']:
for replica in rfile['rses'][rse]:
replicas.append(replica)
dictreplica[replica] = rse
# sort the actual replicas if necessary
if select == 'geoip':
try:
replicas = geoIP_order(dictreplica, client_ip)
except AddressNotFoundError:
pass
else:
replicas = random_order(dictreplica, client_ip)
# stream metadata
yield ' <file name="' + rfile['name'] + '">\n'
yield ' <identity>' + rfile['scope'] + ':' + rfile[
'name'] + '</identity>\n'
if rfile['adler32'] is not None:
yield ' <hash type="adler32">' + rfile[
'adler32'] + '</hash>\n'
if rfile['md5'] is not None:
yield ' <hash type="md5">' + rfile['md5'] + '</hash>\n'
yield ' <size>' + str(rfile['bytes']) + '</size>\n'
yield ' <glfn name="%s%s">' % (
'root://atlas-xrd-eu.cern.ch:1094//atlas/rucio/', '%s:%s' %
(rfile['scope'], rfile['name']))
yield '</glfn>\n'
# stream URLs
idx = 1
for replica in replicas:
yield ' <url location="' + str(
dictreplica[replica]) + '" priority="' + str(
idx) + '">' + replica + '</url>\n'
idx += 1
yield ' </file>\n'
# don't forget to send the metalink footer
yield '</metalink>\n'
except DataIdentifierNotFound, e:
raise generate_http_error(404, 'DataIdentifierNotFound',
e.args[0][0])
def GET(self, scope, name):
"""
Header Redirect
HTTP Success:
303 See Other
HTTP Error:
401 Unauthorized
500 InternalError
404 Notfound
:param scope: The scope name of the file.
:param name: The name of the file.
"""
header('Access-Control-Allow-Origin', ctx.env.get('HTTP_ORIGIN'))
header('Access-Control-Allow-Headers',
ctx.env.get('HTTP_ACCESS_CONTROL_REQUEST_HEADERS'))
header('Access-Control-Allow-Methods', '*')
header('Access-Control-Allow-Credentials', 'true')
try:
# use the default HTTP protocols if no scheme is given
select, rse, site, schemes = 'random', None, None, [
'http', 'https', 's3+rucio'
]
if ctx.query:
params = parse_qs(ctx.query[1:])
if 'select' in params:
select = params['select'][0]
if 'rse' in params:
rse = params['rse'][0]
if 'site' in params:
site = params['site'][0]
if 'schemes' in params:
schemes = params['schemes'][0]
# correctly forward the schemes and select to potential metalink followups
cleaned_url = ctx.env.get('REQUEST_URI').split('?')[0]
if isinstance(schemes, list):
header(
'Link',
'<%s/metalink?schemes=%s&select=%s>; rel=describedby; type="application/metalink+xml"'
% (cleaned_url, ','.join(schemes), select))
else:
header(
'Link',
'<%s/metalink?schemes=%s&select=%s>; rel=describedby; type="application/metalink+xml"'
% (cleaned_url, schemes, select))
schemes = [schemes] # list_replicas needs a list
replicas = [
r for r in list_replicas(dids=[{
'scope': scope,
'name': name,
'type': 'FILE'
}],
schemes=schemes)
]
selected_url, selected_rse = None, None
for r in replicas:
if r['rses']:
replicadict = {}
if rse:
if rse in r['rses'] and r['rses'][rse]:
selected_url = r['rses'][rse][0]
selected_rse = rse
else:
raise ReplicaNotFound(
'no redirection possible - no valid RSE for HTTP redirection found'
)
else:
for rep in r['rses']:
for replica in r['rses'][rep]:
replicadict[replica] = rep
if not replicadict:
raise ReplicaNotFound(
'no redirection possible - no valid RSE for HTTP redirection found'
)
elif site:
rep = site_selector(replicadict, site)
if rep:
selected_url = rep[0]
else:
raise ReplicaNotFound(
'no redirection possible - no valid RSE for HTTP redirection found'
)
else:
client_ip = ctx.env.get('HTTP_X_FORWARDED_FOR')
if client_ip is None:
client_ip = ctx.ip
if select == 'geoip':
rep = geoIP_order(replicadict, client_ip)
else:
rep = random_order(replicadict, client_ip)
selected_url = rep[0]
for rep in r['rses']:
for replica in r['rses'][rep]:
if selected_url == replica:
selected_rse = rep
if selected_url:
if selected_url.startswith('s3+rucio://'):
connect(selected_rse, selected_url)
signed_URLS = get_signed_urls([selected_url],
rse=selected_rse,
operation='read')
raise seeother(signed_URLS[selected_url])
raise seeother(selected_url)
raise ReplicaNotFound(
'no redirection possible - file does not exist')
except seeother:
raise
except ReplicaNotFound, e:
raise generate_http_error(404, 'ReplicaNotFound', e.args[0][0])
