def _configure_arakoon_to_volumedriver():
print 'Update existing vPools'
logger.info('Update existing vPools')
config = ArakoonClusterConfig('voldrv')
config.load_config()
arakoon_nodes = []
for node in config.nodes:
arakoon_nodes.append({
'host': node.ip,
'port': node.client_port,
'node_id': node.name
})
if EtcdConfiguration.dir_exists('/ovs/vpools'):
for vpool_guid in EtcdConfiguration.list('/ovs/vpools'):
for storagedriver_id in EtcdConfiguration.list(
'/ovs/vpools/{0}/hosts'.format(vpool_guid)):
storagedriver_config = StorageDriverConfiguration(
'storagedriver', vpool_guid, storagedriver_id)
storagedriver_config.load()
storagedriver_config.configure_volume_registry(
vregistry_arakoon_cluster_id='voldrv',
vregistry_arakoon_cluster_nodes=arakoon_nodes)
storagedriver_config.configure_distributed_lock_store(
dls_type='Arakoon',
dls_arakoon_cluster_id='voldrv',
dls_arakoon_cluster_nodes=arakoon_nodes)
storagedriver_config.save(reload_config=True)
def get_unused_arakoon_metadata_and_claim(cluster_type, locked=True):
"""
Retrieve arakoon cluster information based on its type
:param cluster_type: Type of arakoon cluster (See ServiceType.ARAKOON_CLUSTER_TYPES)
:type cluster_type: str
:param locked: Execute this in a locked context
:type locked: bool
:return: List of ArakoonClusterMetadata objects
:rtype: ArakoonClusterMetadata
"""
cluster_type = cluster_type.upper()
if cluster_type not in ServiceType.ARAKOON_CLUSTER_TYPES:
raise ValueError('Unsupported arakoon cluster type provided. Please choose from {0}'.format(', '.join(ServiceType.ARAKOON_CLUSTER_TYPES)))
if not EtcdConfiguration.dir_exists('/ovs/arakoon'):
return None
mutex = volatile_mutex('claim_arakoon_metadata', wait=10)
try:
if locked is True:
mutex.acquire()
for cluster_name in EtcdConfiguration.list('/ovs/arakoon'):
metadata = ArakoonClusterMetadata(cluster_id=cluster_name)
metadata.load_metadata()
if metadata.cluster_type == cluster_type and metadata.in_use is False and metadata.internal is False:
metadata.claim()
return metadata
finally:
if locked is True:
mutex.release()
def get_unused_arakoon_metadata_and_claim(cluster_type, locked=True):
"""
Retrieve arakoon cluster information based on its type
:param cluster_type: Type of arakoon cluster (See ServiceType.ARAKOON_CLUSTER_TYPES)
:type cluster_type: str
:param locked: Execute this in a locked context
:type locked: bool
:return: List of ArakoonClusterMetadata objects
:rtype: ArakoonClusterMetadata
"""
cluster_type = cluster_type.upper()
if cluster_type not in ServiceType.ARAKOON_CLUSTER_TYPES:
raise ValueError('Unsupported arakoon cluster type provided. Please choose from {0}'.format(', '.join(ServiceType.ARAKOON_CLUSTER_TYPES)))
if not EtcdConfiguration.dir_exists('/ovs/arakoon'):
return None
mutex = volatile_mutex('claim_arakoon_metadata', wait=10)
try:
if locked is True:
mutex.acquire()
for cluster_name in EtcdConfiguration.list('/ovs/arakoon'):
metadata = ArakoonClusterMetadata(cluster_id=cluster_name)
metadata.load_metadata()
if metadata.cluster_type == cluster_type and metadata.in_use is False and metadata.internal is False:
metadata.claim()
return metadata
finally:
if locked is True:
mutex.release()
def _get_free_ports(client):
node_name = System.get_my_machine_id(client)
clusters = []
exclude_ports = []
if EtcdConfiguration.dir_exists(ArakoonInstaller.ETCD_CONFIG_ROOT):
for cluster_name in EtcdConfiguration.list(
ArakoonInstaller.ETCD_CONFIG_ROOT):
try:
config = ArakoonClusterConfig(cluster_name)
config.load_config()
for node in config.nodes:
if node.name == node_name:
clusters.append(cluster_name)
exclude_ports.append(node.client_port)
exclude_ports.append(node.messaging_port)
except:
logger.error(
' Could not load port information of cluster {0}'.
format(cluster_name))
ports = System.get_free_ports(
EtcdConfiguration.get(
'/ovs/framework/hosts/{0}/ports|arakoon'.format(node_name)),
exclude_ports, 2, client)
logger.debug(
' Loaded free ports {0} based on existing clusters {1}'.format(
ports, clusters))
return ports
def list(self, discover=False, ip=None, node_id=None):
"""
Lists all available ALBA Nodes
:param discover: If True and IP provided, return list of single ALBA node, If True and no IP provided, return all ALBA nodes else return modeled ALBA nodes
:param ip: IP of ALBA node to retrieve
:param node_id: ID of the ALBA node
"""
if discover is False and (ip is not None or node_id is not None):
raise RuntimeError('Discover is mutually exclusive with IP and nodeID')
if (ip is None and node_id is not None) or (ip is not None and node_id is None):
raise RuntimeError('Both IP and nodeID need to be specified')
if discover is False:
return AlbaNodeList.get_albanodes()
if ip is not None:
node = AlbaNode(volatile=True)
node.ip = ip
node.type = 'ASD'
node.node_id = node_id
node.port = EtcdConfiguration.get('/ovs/alba/asdnodes/{0}/config/main|port'.format(node_id))
node.username = EtcdConfiguration.get('/ovs/alba/asdnodes/{0}/config/main|username'.format(node_id))
node.password = EtcdConfiguration.get('/ovs/alba/asdnodes/{0}/config/main|password'.format(node_id))
data = node.client.get_metadata()
if data['_success'] is False and data['_error'] == 'Invalid credentials':
raise RuntimeError('Invalid credentials')
if data['node_id'] != node_id:
raise RuntimeError('Unexpected node identifier. {0} vs {1}'.format(data['node_id'], node_id))
node_list = DataList(AlbaNode, {})
node_list._executed = True
node_list._guids = [node.guid]
node_list._objects = {node.guid: node}
node_list._data = {node.guid: {'guid': node.guid, 'data': node._data}}
return node_list
nodes = {}
model_node_ids = [node.node_id for node in AlbaNodeList.get_albanodes()]
found_node_ids = []
asd_node_ids = []
if EtcdConfiguration.dir_exists('/ovs/alba/asdnodes'):
asd_node_ids = EtcdConfiguration.list('/ovs/alba/asdnodes')
for node_id in asd_node_ids:
node = AlbaNode(volatile=True)
node.type = 'ASD'
node.node_id = node_id
node.ip = EtcdConfiguration.get('/ovs/alba/asdnodes/{0}/config/main|ip'.format(node_id))
node.port = EtcdConfiguration.get('/ovs/alba/asdnodes/{0}/config/main|port'.format(node_id))
node.username = EtcdConfiguration.get('/ovs/alba/asdnodes/{0}/config/main|username'.format(node_id))
node.password = EtcdConfiguration.get('/ovs/alba/asdnodes/{0}/config/main|password'.format(node_id))
if node.node_id not in model_node_ids and node.node_id not in found_node_ids:
nodes[node.guid] = node
found_node_ids.append(node.node_id)
node_list = DataList(AlbaNode, {})
node_list._executed = True
node_list._guids = nodes.keys()
node_list._objects = nodes
node_list._data = dict([(node.guid, {'guid': node.guid, 'data': node._data}) for node in nodes.values()])
return node_list
def validate_alba_backend_removal(alba_backend_info):
"""
Validate whether the backend has been deleted properly
alba_backend_info should be a dictionary containing:
- guid
- name
- maintenance_service_names
:param alba_backend_info: Information about the backend
:return: None
"""
Toolbox.verify_required_params(actual_params=alba_backend_info,
required_params={'name': (str, None),
'guid': (str, Toolbox.regex_guid),
'maintenance_service_names': (list, None)},
exact_match=True)
alba_backend_guid = alba_backend_info['guid']
alba_backend_name = alba_backend_info['name']
backend = GeneralBackend.get_by_name(alba_backend_name)
assert backend is None, 'Still found a backend in the model with name {0}'.format(alba_backend_name)
# Validate services removed from model
for service in GeneralService.get_services_by_name(ServiceType.SERVICE_TYPES.ALBA_MGR):
assert service.name != '{0}-abm'.format(alba_backend_name), 'An AlbaManager service has been found with name {0}'.format(alba_backend_name)
for service in GeneralService.get_services_by_name(ServiceType.SERVICE_TYPES.NS_MGR):
assert service.name.startswith('{0}-nsm_'.format(alba_backend_name)) is False, 'An NamespaceManager service has been found with name {0}'.format(alba_backend_name)
# Validate ALBA backend ETCD structure
alba_backend_key = '/ovs/alba/backends'
actual_etcd_keys = [key for key in EtcdConfiguration.list(alba_backend_key)]
assert alba_backend_guid not in actual_etcd_keys, 'Etcd still contains an entry in {0} with guid {1}'.format(alba_backend_key, alba_backend_guid)
# Validate Arakoon ETCD structure
arakoon_keys = [key for key in EtcdConfiguration.list('/ovs/arakoon') if key.startswith(alba_backend_name)]
assert len(arakoon_keys) == 0, 'Etcd still contains configurations for clusters: {0}'.format(', '.join(arakoon_keys))
# Validate services
for storagerouter in GeneralStorageRouter.get_storage_routers():
root_client = SSHClient(endpoint=storagerouter,
username='******')
maintenance_services = alba_backend_info['maintenance_service_names']
abm_arakoon_service_name = 'ovs-arakoon-{0}-abm'.format(alba_backend_name)
nsm_arakoon_service_name = 'ovs-arakoon-{0}-nsm_0'.format(alba_backend_name)
for service_name in [abm_arakoon_service_name, nsm_arakoon_service_name] + maintenance_services:
assert GeneralService.has_service(name=service_name,
client=root_client) is False, 'Service {0} still deployed on Storage Router {1}'.format(service_name, storagerouter.name)
def _configure_arakoon_to_volumedriver(cluster_name):
print 'Update existing vPools'
StorageDriverController._logger.info('Update existing vPools')
config = ArakoonClusterConfig(cluster_name)
config.load_config()
arakoon_nodes = []
for node in config.nodes:
arakoon_nodes.append({'host': node.ip,
'port': node.client_port,
'node_id': node.name})
if EtcdConfiguration.dir_exists('/ovs/vpools'):
for vpool_guid in EtcdConfiguration.list('/ovs/vpools'):
for storagedriver_id in EtcdConfiguration.list('/ovs/vpools/{0}/hosts'.format(vpool_guid)):
storagedriver_config = StorageDriverConfiguration('storagedriver', vpool_guid, storagedriver_id)
storagedriver_config.load()
storagedriver_config.configure_volume_registry(vregistry_arakoon_cluster_id=cluster_name,
vregistry_arakoon_cluster_nodes=arakoon_nodes)
storagedriver_config.configure_distributed_lock_store(dls_type='Arakoon',
dls_arakoon_cluster_id=cluster_name,
dls_arakoon_cluster_nodes=arakoon_nodes)
storagedriver_config.save(reload_config=True)
def get_arakoon_metadata_by_cluster_name(cluster_name):
"""
Retrieve arakoon cluster information based on its name
:param cluster_name: Name of the arakoon cluster
:type cluster_name: str
:return: Arakoon cluster metadata information
:rtype: ArakoonClusterMetadata
"""
if not EtcdConfiguration.exists('/ovs/arakoon', raw=True):
raise ValueError('Etcd key "/ovs/arakoon" not found')
for cluster in EtcdConfiguration.list('/ovs/arakoon'):
if cluster == cluster_name:
arakoon_metadata = ArakoonClusterMetadata(cluster_id=cluster_name)
arakoon_metadata.load_metadata()
return arakoon_metadata
raise ValueError('No arakoon cluster found with name "{0}"'.format(cluster_name))
def get_arakoon_metadata_by_cluster_name(cluster_name):
"""
Retrieve arakoon cluster information based on its name
:param cluster_name: Name of the arakoon cluster
:type cluster_name: str
:return: Arakoon cluster metadata information
:rtype: ArakoonClusterMetadata
"""
if not EtcdConfiguration.exists('/ovs/arakoon', raw=True):
raise ValueError('Etcd key "/ovs/arakoon" not found')
for cluster in EtcdConfiguration.list('/ovs/arakoon'):
if cluster == cluster_name:
arakoon_metadata = ArakoonClusterMetadata(cluster_id=cluster_name)
arakoon_metadata.load_metadata()
return arakoon_metadata
raise ValueError('No arakoon cluster found with name "{0}"'.format(cluster_name))
def _get_free_ports(client):
node_name = System.get_my_machine_id(client)
clusters = []
exclude_ports = []
if EtcdConfiguration.dir_exists(ArakoonInstaller.ETCD_CONFIG_ROOT):
for cluster_name in EtcdConfiguration.list(ArakoonInstaller.ETCD_CONFIG_ROOT):
try:
config = ArakoonClusterConfig(cluster_name)
config.load_config()
for node in config.nodes:
if node.name == node_name:
clusters.append(cluster_name)
exclude_ports.append(node.client_port)
exclude_ports.append(node.messaging_port)
except:
logger.error(' Could not load port information of cluster {0}'.format(cluster_name))
ports = System.get_free_ports(EtcdConfiguration.get('/ovs/framework/hosts/{0}/ports|arakoon'.format(node_name)), exclude_ports, 2, client)
logger.debug(' Loaded free ports {0} based on existing clusters {1}'.format(ports, clusters))
return ports
def model_local_albanode(**kwargs):
"""
Add all ALBA nodes known to etcd to the model
:param kwargs: Kwargs containing information regarding the node
:type kwargs: dict
:return: None
"""
_ = kwargs
if EtcdConfiguration.dir_exists('/ovs/alba/asdnodes'):
for node_id in EtcdConfiguration.list('/ovs/alba/asdnodes'):
node = AlbaNodeList.get_albanode_by_node_id(node_id)
if node is None:
node = AlbaNode()
main_config = EtcdConfiguration.get('/ovs/alba/asdnodes/{0}/config/main'.format(node_id))
node.type = 'ASD'
node.node_id = node_id
node.ip = main_config['ip']
node.port = main_config['port']
node.username = main_config['username']
node.password = main_config['password']
node.storagerouter = StorageRouterList.get_by_ip(main_config['ip'])
node.save()
def check_vpool_cleanup(vpool_info, storagerouters=None):
"""
Check if everything related to a vPool has been cleaned up on the storagerouters provided
vpool_info should be a dictionary containing:
- type
- guid
- files
- directories
- name (optional)
- vpool (optional)
If vpool is provided:
- storagerouters need to be provided, because on these Storage Routers, we check whether the vPool has been cleaned up
If name is provided:
- If storagerouters is NOT provided, all Storage Routers will be checked for a correct vPool removal
- If storagerouters is provided, only these Storage Routers will be checked for a correct vPool removal
:param vpool_info: Information about the vPool
:param storagerouters: Storage Routers to check if vPool has been cleaned up
:return: None
"""
for required_param in ['type', 'guid', 'files', 'directories']:
if required_param not in vpool_info:
raise ValueError('Incorrect vpool_info provided')
if 'vpool' in vpool_info and 'name' in vpool_info:
raise ValueError('vpool and name are mutually exclusive')
if 'vpool' not in vpool_info and 'name' not in vpool_info:
raise ValueError('Either vpool or vpool_name needs to be provided')
vpool = vpool_info.get('vpool')
vpool_name = vpool_info.get('name')
vpool_guid = vpool_info['guid']
vpool_type = vpool_info['type']
files = vpool_info['files']
directories = vpool_info['directories']
supported_backend_types = GeneralBackend.get_valid_backendtypes()
if vpool_type not in supported_backend_types:
raise ValueError('Unsupported Backend Type provided. Please choose from: {0}'.format(', '.join(supported_backend_types)))
if storagerouters is None:
storagerouters = GeneralStorageRouter.get_storage_routers()
if vpool_name is not None:
assert GeneralVPool.get_vpool_by_name(vpool_name=vpool_name) is None, 'A vPool with name {0} still exists'.format(vpool_name)
# Prepare some fields to check
vpool_name = vpool.name if vpool else vpool_name
vpool_services = ['ovs-dtl_{0}'.format(vpool_name),
'ovs-volumedriver_{0}'.format(vpool_name)]
if vpool_type == 'alba':
vpool_services.append('ovs-albaproxy_{0}'.format(vpool_name))
# Check etcd
if vpool is None:
assert EtcdConfiguration.exists('/ovs/vpools/{0}'.format(vpool_guid), raw=True) is False, 'vPool config still found in etcd'
else:
remaining_sd_ids = set([storagedriver.storagedriver_id for storagedriver in vpool.storagedrivers])
current_sd_ids = set([item for item in EtcdConfiguration.list('/ovs/vpools/{0}/hosts'.format(vpool_guid))])
assert not remaining_sd_ids.difference(current_sd_ids), 'There are more storagedrivers modelled than present in etcd'
assert not current_sd_ids.difference(remaining_sd_ids), 'There are more storagedrivers in etcd than present in model'
# Perform checks on all storagerouters where vpool was removed
for storagerouter in storagerouters:
# Check management center
mgmt_center = GeneralManagementCenter.get_mgmt_center(pmachine=storagerouter.pmachine)
if mgmt_center is not None:
assert GeneralManagementCenter.is_host_configured(pmachine=storagerouter.pmachine) is False, 'Management Center is still configured on Storage Router {0}'.format(storagerouter.ip)
# Check MDS services
mds_services = GeneralService.get_services_by_name(ServiceType.SERVICE_TYPES.MD_SERVER)
assert len([mds_service for mds_service in mds_services if mds_service.storagerouter_guid == storagerouter.guid]) == 0, 'There are still MDS services present for Storage Router {0}'.format(storagerouter.ip)
# Check services
root_client = SSHClient(storagerouter, username='******')
for service in vpool_services:
if ServiceManager.has_service(service, client=root_client):
raise RuntimeError('Service {0} is still configured on Storage Router {1}'.format(service, storagerouter.ip))
# Check KVM vpool
if storagerouter.pmachine.hvtype == 'KVM':
vpool_overview = root_client.run('virsh pool-list --all').splitlines()
vpool_overview.pop(1)
vpool_overview.pop(0)
for vpool_info in vpool_overview:
kvm_vpool_name = vpool_info.split()[0].strip()
if vpool_name == kvm_vpool_name:
raise ValueError('vPool {0} is still defined on Storage Router {1}'.format(vpool_name, storagerouter.ip))
# Check file and directory existence
if storagerouter.guid not in directories:
raise ValueError('Could not find directory information for Storage Router {0}'.format(storagerouter.ip))
if storagerouter.guid not in files:
raise ValueError('Could not find file information for Storage Router {0}'.format(storagerouter.ip))
for directory in directories[storagerouter.guid]:
assert root_client.dir_exists(directory) is False, 'Directory {0} still exists on Storage Router {1}'.format(directory, storagerouter.ip)
for file_name in files[storagerouter.guid]:
assert root_client.file_exists(file_name) is False, 'File {0} still exists on Storage Router {1}'.format(file_name, storagerouter.ip)
# Look for errors in storagedriver log
for error_type in ['error', 'fatal']:
cmd = "cat -vet /var/log/ovs/volumedriver/{0}.log | tail -1000 | grep ' {1} '; echo true > /dev/null".format(vpool_name, error_type)
errors = []
for line in root_client.run(cmd).splitlines():
if "HierarchicalArakoon" in line:
continue
errors.append(line)
if len(errors) > 0:
if error_type == 'error':
print 'Volumedriver log file contains errors on Storage Router {0}\n - {1}'.format(storagerouter.ip, '\n - '.join(errors))
else:
raise RuntimeError('Fatal errors found in volumedriver log file on Storage Router {0}\n - {1}'.format(storagerouter.ip, '\n - '.join(errors)))
def validate_vpool_sanity(expected_settings):
"""
Check if all requirements are met for a healthy vPool
:param expected_settings: Parameters used to create a vPool, which will be verified
:type expected_settings: dict
:return: None
"""
if not isinstance(expected_settings, dict) or len(expected_settings) == 0:
raise ValueError('Cannot validate vpool when no settings are passed')
generic_settings = expected_settings.values()[0]
vpool_name = generic_settings['vpool_name']
mountpoint = '/mnt/{0}'.format(vpool_name)
backend_type = generic_settings['type']
rdma_enabled = generic_settings['config_params']['dtl_transport'] == StorageDriverClient.FRAMEWORK_DTL_TRANSPORT_RSOCKET
vpool = GeneralVPool.get_vpool_by_name(vpool_name=vpool_name)
assert vpool is not None, 'Could not find vPool with name {0}'.format(vpool_name)
vpool_config = GeneralVPool.get_configuration(vpool)
# Verify some basic vPool attributes
assert vpool.name == vpool_name, 'Expected name {0} for vPool'.format(vpool_name)
assert vpool.backend_type.code == backend_type, 'Expected backend type {0}'.format(backend_type)
assert vpool.status == VPool.STATUSES.RUNNING, 'vPool does not have RUNNING status'
assert vpool.rdma_enabled == rdma_enabled, 'RDMA enabled setting is incorrect'
assert set(expected_settings.keys()) == set([sd.storagerouter for sd in vpool.storagedrivers]), "vPool storagerouters don't match the expected Storage Routers"
# Verify vPool Storage Driver configuration
expected_vpool_config = copy.deepcopy(generic_settings['config_params'])
for key, value in vpool_config.iteritems():
if key == 'dtl_enabled' or key == 'tlog_multiplier':
continue
if key not in expected_vpool_config:
raise ValueError('Expected settings does not contain key {0}'.format(key))
if value != expected_vpool_config[key]:
raise ValueError('vPool does not have expected configuration {0} for key {1}'.format(expected_vpool_config[key], key))
expected_vpool_config.pop(key)
if len(expected_vpool_config) > 0:
raise ValueError('Actual vPool configuration does not contain keys: {0}'.format(', '.join(expected_vpool_config.keys())))
# Prepare some fields to check
config = generic_settings['config_params']
dtl_mode = config['dtl_mode']
sco_size = config['sco_size']
dedupe_mode = config['dedupe_mode']
cluster_size = config['cluster_size']
write_buffer = config['write_buffer']
dtl_transport = config['dtl_transport']
cache_strategy = config['cache_strategy']
# @TODO: Add more validations for other expected settings (instead of None)
expected_config = {'backend_connection_manager': {'backend_interface_retries_on_error': 5,
'backend_interface_retry_interval_secs': 1,
'backend_interface_retry_backoff_multiplier': 2.0},
'content_addressed_cache': {'clustercache_mount_points': None,
'read_cache_serialization_path': u'/var/rsp/{0}'.format(vpool.name)},
'distributed_lock_store': {'dls_arakoon_cluster_id': None,
'dls_arakoon_cluster_nodes': None,
'dls_type': u'Arakoon'},
'distributed_transaction_log': {'dtl_path': None,
'dtl_transport': dtl_transport.upper()},
'event_publisher': {'events_amqp_routing_key': u'volumerouter',
'events_amqp_uris': None},
'file_driver': {'fd_cache_path': None,
'fd_extent_cache_capacity': u'1024',
'fd_namespace': None},
'filesystem': {'fs_dtl_config_mode': u'Automatic',
'fs_dtl_mode': u'{0}'.format(StorageDriverClient.VPOOL_DTL_MODE_MAP[dtl_mode]),
'fs_enable_shm_interface': 1,
'fs_file_event_rules': None,
'fs_metadata_backend_arakoon_cluster_nodes': None,
'fs_metadata_backend_mds_nodes': None,
'fs_metadata_backend_type': u'MDS',
'fs_raw_disk_suffix': None,
'fs_virtual_disk_format': None},
'metadata_server': {'mds_nodes': None},
'scocache': {'backoff_gap': u'2GB',
'scocache_mount_points': None,
'trigger_gap': u'1GB'},
'threadpool_component': {'num_threads': 16},
'volume_manager': {'clean_interval': 1,
'default_cluster_size': 1024 * cluster_size,
'dtl_throttle_usecs': 4000,
'metadata_path': None,
'non_disposable_scos_factor': float(write_buffer) / StorageDriverClient.TLOG_MULTIPLIER_MAP[sco_size] / sco_size,
'number_of_scos_in_tlog': StorageDriverClient.TLOG_MULTIPLIER_MAP[sco_size],
'read_cache_default_behaviour': StorageDriverClient.VPOOL_CACHE_MAP[cache_strategy],
'read_cache_default_mode': StorageDriverClient.VPOOL_DEDUPE_MAP[dedupe_mode],
'tlog_path': None},
'volume_registry': {'vregistry_arakoon_cluster_id': u'voldrv',
'vregistry_arakoon_cluster_nodes': None},
'volume_router': {'vrouter_backend_sync_timeout_ms': 5000,
'vrouter_file_read_threshold': 1024,
'vrouter_file_write_threshold': 1024,
'vrouter_id': None,
'vrouter_max_workers': 16,
'vrouter_migrate_timeout_ms': 5000,
'vrouter_min_workers': 4,
'vrouter_redirect_timeout_ms': u'5000',
'vrouter_routing_retries': 10,
'vrouter_sco_multiplier': 1024,
'vrouter_volume_read_threshold': 1024,
'vrouter_volume_write_threshold': 1024},
'volume_router_cluster': {'vrouter_cluster_id': None}}
vpool_services = {'all': ['ovs-watcher-volumedriver',
'ovs-dtl_{0}'.format(vpool.name),
'ovs-volumedriver_{0}'.format(vpool.name),
'ovs-volumerouter-consumer'],
'extra': [],
'master': ['ovs-arakoon-voldrv']}
sd_partitions = {'DB': ['MD', 'MDS', 'TLOG'],
'READ': ['None'],
'WRITE': ['FD', 'DTL', 'SCO'],
'SCRUB': ['None']}
if backend_type == 'alba':
backend_metadata = {'name': (str, None),
'preset': (str, Toolbox.regex_preset),
'backend_guid': (str, Toolbox.regex_guid),
'arakoon_config': (dict, None),
'connection': (dict, {'host': (str, Toolbox.regex_ip, False),
'port': (int, {'min': 1, 'max': 65535}),
'client_id': (str, Toolbox.regex_guid),
'client_secret': (str, None),
'local': (bool, None)}),
'backend_info': (dict, {'policies': (list, None),
'sco_size': (float, None),
'frag_size': (float, None),
'total_size': (float, None),
'nsm_partition_guids': (list, Toolbox.regex_guid)})}
required = {'backend': (dict, backend_metadata),
'backend_aa': (dict, backend_metadata, False)}
Toolbox.verify_required_params(required_params=required,
actual_params=vpool.metadata)
vpool_services['all'].append("ovs-albaproxy_{0}".format(vpool.name))
sd_partitions['WRITE'].append('FCACHE')
expected_config['backend_connection_manager'].update({'alba_connection_host': None,
'alba_connection_port': None,
'alba_connection_preset': None,
'alba_connection_timeout': 15,
'backend_type': u'{0}'.format(vpool.backend_type.code.upper())})
elif backend_type == 'distributed':
expected_config['backend_connection_manager'].update({'backend_type': u'LOCAL',
'local_connection_path': u'{0}'.format(generic_settings['distributed_mountpoint'])})
assert EtcdConfiguration.exists('/ovs/arakoon/voldrv/config', raw=True), 'Volumedriver arakoon does not exist'
# Do some verifications for all SDs
storage_ip = None
voldrv_config = GeneralArakoon.get_config('voldrv')
all_files = GeneralVPool.get_related_files(vpool=vpool)
all_directories = GeneralVPool.get_related_directories(vpool=vpool)
for storagedriver in vpool.storagedrivers:
storagerouter = storagedriver.storagerouter
root_client = SSHClient(storagerouter, username='******')
assert EtcdConfiguration.exists('/ovs/vpools/{0}/hosts/{1}/config'.format(vpool.guid, storagedriver.storagedriver_id), raw=True), 'vPool config not found in etcd'
current_config_sections = set([item for item in EtcdConfiguration.list('/ovs/vpools/{0}/hosts/{1}/config'.format(vpool.guid, storagedriver.storagedriver_id))])
assert not current_config_sections.difference(set(expected_config.keys())), 'New section appeared in the storage driver config in etcd'
assert not set(expected_config.keys()).difference(current_config_sections), 'Config section expected for storage driver, but not found in etcd'
for key, values in expected_config.iteritems():
current_config = EtcdConfiguration.get('/ovs/vpools/{0}/hosts/{1}/config/{2}'.format(vpool.guid, storagedriver.storagedriver_id, key))
assert set(current_config.keys()).union(set(values.keys())) == set(values.keys()), 'Not all expected keys match for key "{0}" on Storage Driver {1}'.format(key, storagedriver.name)
for sub_key, value in current_config.iteritems():
expected_value = values[sub_key]
if expected_value is None:
continue
assert value == expected_value, 'Key: {0} - Sub key: {1} - Value: {2} - Expected value: {3}'.format(key, sub_key, value, expected_value)
# Check services
if storagerouter.node_type == 'MASTER':
for service_name in vpool_services['all'] + vpool_services['master']:
if service_name == 'ovs-arakoon-voldrv' and GeneralStorageDriver.has_role(storagedriver, 'DB') is False:
continue
if ServiceManager.get_service_status(name=service_name,
client=root_client) is not True:
raise ValueError('Service {0} is not running on node {1}'.format(service_name, storagerouter.ip))
else:
for service_name in vpool_services['all'] + vpool_services['extra']:
if ServiceManager.get_service_status(name=service_name,
client=root_client) is not True:
raise ValueError('Service {0} is not running on node {1}'.format(service_name, storagerouter.ip))
# Check arakoon config
if not voldrv_config.has_section(storagerouter.machine_id):
raise ValueError('Voldrv arakoon cluster does not have section {0}'.format(storagerouter.machine_id))
# Basic SD checks
assert storagedriver.cluster_ip == storagerouter.ip, 'Incorrect cluster IP. Expected: {0} - Actual: {1}'.format(storagerouter.ip, storagedriver.cluster_ip)
assert storagedriver.mountpoint == '/mnt/{0}'.format(vpool.name), 'Incorrect mountpoint. Expected: {0} - Actual: {1}'.format(mountpoint, storagedriver.mountpoint)
if storage_ip is not None:
assert storagedriver.storage_ip == storage_ip, 'Incorrect storage IP. Expected: {0} - Actual: {1}'.format(storage_ip, storagedriver.storage_ip)
storage_ip = storagedriver.storage_ip
# Check required directories and files
if storagerouter.guid not in all_directories:
raise ValueError('Could not find directory information for Storage Router {0}'.format(storagerouter.ip))
if storagerouter.guid not in all_files:
raise ValueError('Could not find file information for Storage Router {0}'.format(storagerouter.ip))
for directory in all_directories[storagerouter.guid]:
if root_client.dir_exists(directory) is False:
raise ValueError('Directory {0} does not exist on Storage Router {1}'.format(directory, storagerouter.ip))
for file_name in all_files[storagerouter.guid]:
if root_client.file_exists(file_name) is False:
raise ValueError('File {0} does not exist on Storage Router {1}'.format(file_name, storagerouter.ip))
for partition in storagedriver.partitions:
if partition.role in sd_partitions and partition.sub_role in sd_partitions[partition.role]:
sd_partitions[partition.role].remove(partition.sub_role)
elif partition.role in sd_partitions and partition.sub_role is None:
sd_partitions[partition.role].remove('None')
# Verify vPool writeable
if storagerouter.pmachine.hvtype == 'VMWARE':
GeneralVPool.mount_vpool(vpool=vpool,
root_client=root_client)
vdisk = GeneralVDisk.create_volume(size=10,
vpool=vpool,
root_client=root_client)
GeneralVDisk.write_to_volume(vdisk=vdisk,
vpool=vpool,
root_client=root_client,
count=10,
bs='1M',
input_type='random')
GeneralVDisk.delete_volume(vdisk=vdisk,
vpool=vpool,
root_client=root_client)
for role, sub_roles in sd_partitions.iteritems():
for sub_role in sub_roles:
raise ValueError('Not a single Storage Driver found with partition role {0} and sub-role {1}'.format(role, sub_role))
def validate_alba_backend_sanity_without_claimed_disks(alba_backend):
"""
Validate whether the ALBA backend is configured correctly
:param alba_backend: ALBA backend
:return: None
"""
# Attribute validation
assert alba_backend.available is True, 'ALBA backend {0} is not available'.format(alba_backend.backend.name)
assert len(alba_backend.presets) >= 1, 'No preset found for ALBA backend {0}'.format(alba_backend.backend.name)
assert len([default for default in alba_backend.presets if default['is_default'] is True]) == 1, 'Could not find default preset for backend {0}'.format(alba_backend.backend.name)
assert alba_backend.backend.backend_type.code == 'alba', 'Backend type for ALBA backend is {0}'.format(alba_backend.backend.backend_type.code)
assert alba_backend.backend.status == 'RUNNING', 'Status for ALBA backend is {0}'.format(alba_backend.backend.status)
assert isinstance(alba_backend.metadata_information, dict) is True, 'ALBA backend {0} metadata information is not a dictionary'.format(alba_backend.backend.name)
Toolbox.verify_required_params(actual_params=alba_backend.metadata_information,
required_params={'nsm_partition_guids': (list, Toolbox.regex_guid)},
exact_match=True)
# Validate ABM and NSM services
storagerouters = GeneralStorageRouter.get_storage_routers()
storagerouters_with_db_role = [sr for sr in storagerouters if GeneralStorageRouter.has_roles(storagerouter=sr, roles='DB') is True and sr.node_type == 'MASTER']
assert len(alba_backend.abm_services) == len(storagerouters_with_db_role), 'Not enough ABM services found'
assert len(alba_backend.nsm_services) == len(storagerouters_with_db_role), 'Not enough NSM services found'
# Validate ALBA backend ETCD structure
alba_backend_key = '/ovs/alba/backends'
assert EtcdConfiguration.exists(key=alba_backend_key, raw=True) is True, 'Etcd does not contain key {0}'.format(alba_backend_key)
actual_etcd_keys = [key for key in EtcdConfiguration.list(alba_backend_key)]
expected_etcd_keys = ['verification_schedule', 'global_gui_error_interval', alba_backend.guid,
'default_nsm_hosts']
optional_etcd_keys = ['verification_factor']
expected_keys_amount = 0
for optional_key in optional_etcd_keys:
if optional_key in actual_etcd_keys:
expected_keys_amount += 1
for expected_key in expected_etcd_keys:
if not re.match(Toolbox.regex_guid, expected_key):
expected_keys_amount += 1
assert expected_key in actual_etcd_keys, 'Key {0} was not found in tree {1}'.format(expected_key, alba_backend_key)
for actual_key in list(actual_etcd_keys):
if re.match(Toolbox.regex_guid, actual_key):
actual_etcd_keys.remove(actual_key) # Remove all alba backend keys
assert len(actual_etcd_keys) == expected_keys_amount, 'Another key was added to the {0} tree'.format(alba_backend_key)
this_alba_backend_key = '{0}/{1}'.format(alba_backend_key, alba_backend.guid)
actual_keys = [key for key in EtcdConfiguration.list(this_alba_backend_key)]
expected_keys = ['maintenance']
assert actual_keys == expected_keys, 'Actual keys: {0} - Expected keys: {1}'.format(actual_keys, expected_keys)
maintenance_key = '{0}/maintenance'.format(this_alba_backend_key)
actual_keys = [key for key in EtcdConfiguration.list(maintenance_key)]
expected_keys = ['nr_of_agents', 'config']
assert set(actual_keys) == set(expected_keys), 'Actual keys: {0} - Expected keys: {1}'.format(actual_keys, expected_keys)
# @TODO: Add validation for config values
# Validate ASD node ETCD structure
alba_nodes = GeneralAlba.get_alba_nodes()
assert len(alba_nodes) > 0, 'Could not find any ALBA nodes in the model'
alba_node_key = '/ovs/alba/asdnodes'
actual_keys = [key for key in EtcdConfiguration.list(alba_node_key)]
assert len(alba_nodes) == len(actual_keys), 'Amount of ALBA nodes in model does not match amount of ALBA nodes in ETCD. In model: {0} - In Etcd: {1}'.format(len(alba_nodes), len(actual_keys))
for alba_node in alba_nodes:
assert alba_node.node_id in actual_keys, 'ALBA node with ID {0} not present in ETCD'.format(alba_node.node_id)
actual_asdnode_keys = [key for key in EtcdConfiguration.list('{0}/{1}'.format(alba_node_key, alba_node.node_id))]
expected_asdnode_keys = ['config']
assert actual_asdnode_keys == expected_asdnode_keys, 'Actual keys: {0} - Expected keys: {1}'.format(actual_asdnode_keys, expected_asdnode_keys)
actual_config_keys = [key for key in EtcdConfiguration.list('{0}/{1}/config'.format(alba_node_key, alba_node.node_id))]
expected_config_keys = ['main', 'network']
assert set(actual_config_keys) == set(expected_config_keys), 'Actual keys: {0} - Expected keys: {1}'.format(actual_config_keys, expected_config_keys)
# @TODO: Add validation for main and network values
# Validate Arakoon ETCD structure
arakoon_abm_key = '/ovs/arakoon/{0}/config'.format(alba_backend.abm_services[0].service.name)
arakoon_nsm_key = '/ovs/arakoon/{0}/config'.format(alba_backend.nsm_services[0].service.name)
assert EtcdConfiguration.exists(key=arakoon_abm_key, raw=True) is True, 'Etcd key {0} does not exists'.format(arakoon_abm_key)
assert EtcdConfiguration.exists(key=arakoon_nsm_key, raw=True) is True, 'Etcd key {0} does not exists'.format(arakoon_nsm_key)
# @TODO: Add validation for config values
# Validate maintenance agents
actual_amount_agents = len([service for node_services in [alba_node.client.list_maintenance_services() for alba_node in alba_nodes] for service in node_services])
expected_amount_agents = EtcdConfiguration.get('/ovs/alba/backends/{0}/maintenance/nr_of_agents'.format(alba_backend.guid))
assert actual_amount_agents == expected_amount_agents, 'Amount of maintenance agents is incorrect. Found {0} - Expected {1}'.format(actual_amount_agents, expected_amount_agents)
# Validate arakoon services
machine_ids = [sr.machine_id for sr in storagerouters_with_db_role]
abm_service_name = alba_backend.abm_services[0].service.name
nsm_service_name = alba_backend.nsm_services[0].service.name
for storagerouter in storagerouters_with_db_role:
root_client = SSHClient(endpoint=storagerouter,
username='******')
abm_arakoon_service_name = 'ovs-arakoon-{0}'.format(abm_service_name)
nsm_arakoon_service_name = 'ovs-arakoon-{0}'.format(nsm_service_name)
for service_name in [abm_arakoon_service_name, nsm_arakoon_service_name]:
assert GeneralService.has_service(name=service_name,
client=root_client) is True, 'Service {0} not deployed on Storage Router {1}'.format(service_name, storagerouter.name)
assert GeneralService.get_service_status(name=service_name,
client=root_client) is True, 'Service {0} not running on Storage Router {1}'.format(service_name, storagerouter.name)
out, err, _ = General.execute_command('arakoon --who-master -config {0}'.format(GeneralArakoon.ETCD_CONFIG_PATH.format(abm_service_name)))
assert out.strip() in machine_ids, 'Arakoon master is {0}, but should be 1 of "{1}"'.format(out.strip(), ', '.join(machine_ids))
