def apply(license_string):
"""
Applies a license. It will apply as much licenses as possible, however, it won't fail on invalid licenses as it
will simply skip them.
"""
try:
clients = {}
storagerouters = StorageRouterList.get_storagerouters()
try:
for storagerouter in storagerouters:
clients[storagerouter] = SSHClient(storagerouter.ip)
except UnableToConnectException:
raise RuntimeError('Not all StorageRouters are reachable')
data = LicenseController._decode(license_string)
for component in data:
cdata = data[component]
name = cdata['name']
data = cdata['data']
token = cdata['token']
valid_until = float(
cdata['valid_until']) if 'valid_until' in cdata else None
if valid_until is not None and valid_until <= time.time():
continue
signature = cdata['signature'] if 'signature' in cdata else None
validate_functions = Toolbox.fetch_hooks(
'license', '{0}.validate'.format(component))
apply_functions = Toolbox.fetch_hooks(
'license', '{0}.apply'.format(component))
if len(validate_functions) == 1 and len(apply_functions) == 1:
valid, metadata = validate_functions[0](
component=component, data=data, signature=signature)
if valid is True:
success = apply_functions[0](component=component,
data=data,
signature=signature)
if success is True:
license_object = LicenseList.get_by_component(
component)
if license_object is None:
license_object = License()
license_object.component = component
license_object.name = name
license_object.token = token
license_object.data = data
license_object.valid_until = valid_until
license_object.signature = signature
license_object.save()
license_contents = []
for lic in LicenseList.get_licenses():
license_contents.append(lic.hash)
for storagerouter in storagerouters:
client = clients[storagerouter]
client.file_write('/opt/OpenvStorage/config/licenses',
'{0}\n'.format('\n'.join(license_contents)))
except Exception, ex:
logger.exception('Error applying license: {0}'.format(ex))
return None
def refresh_package_information():
"""
Retrieve and store the package information of all StorageRouters
:return: None
"""
GenericController._logger.info('Updating package information')
threads = []
information = {}
all_storagerouters = StorageRouterList.get_storagerouters()
for storagerouter in all_storagerouters:
information[storagerouter.ip] = {}
for fct in Toolbox.fetch_hooks('update', 'get_package_info_multi'):
try:
# We make use of these clients in Threads --> cached = False
client = SSHClient(endpoint=storagerouter,
username='******',
cached=False)
except UnableToConnectException:
information[storagerouter.ip]['errors'] = [
'StorageRouter {0} is inaccessible'.format(
storagerouter.name)
]
break
thread = Thread(target=fct, args=(client, information))
thread.start()
threads.append(thread)
for fct in Toolbox.fetch_hooks('update', 'get_package_info_single'):
thread = Thread(target=fct, args=(information, ))
thread.start()
threads.append(thread)
for thread in threads:
thread.join()
errors = []
copy_information = copy.deepcopy(information)
for ip, info in information.iteritems():
if len(info.get('errors', [])) > 0:
errors.extend(
['{0}: {1}'.format(ip, error) for error in info['errors']])
copy_information.pop(ip)
for storagerouter in all_storagerouters:
info = copy_information.get(storagerouter.ip, {})
if 'errors' in info:
info.pop('errors')
storagerouter.package_information = info
storagerouter.save()
if len(errors) > 0:
errors = [str(error) for error in set(errors)]
raise Exception(' - {0}'.format('\n - '.join(errors)))
def validate(license_string):
"""
Validates a license with the various components
"""
try:
result = {}
data = LicenseController._decode(license_string)
for component in data:
cdata = data[component]
name = cdata['name']
data = cdata['data']
_ = cdata['token']
valid_until = float(
cdata['valid_until']) if 'valid_until' in cdata else None
if valid_until is not None and valid_until <= time.time():
result[component] = False
continue
signature = cdata['signature'] if 'signature' in cdata else None
validate_functions = Toolbox.fetch_hooks(
'license', '{0}.validate'.format(component))
apply_functions = Toolbox.fetch_hooks(
'license', '{0}.apply'.format(component))
if len(validate_functions) == 1 and len(apply_functions) == 1:
try:
valid, metadata = validate_functions[0](
component=component,
data=data,
signature=signature)
except Exception, ex:
logger.debug(
'Error validating license for {0}: {1}'.format(
component, ex))
valid = False
metadata = None
if valid is False:
logger.debug('Invalid license for {0}: {1}'.format(
component, license_string))
result[component] = False
else:
result[component] = {
'valid_until': valid_until,
'metadata': metadata,
'name': name
}
else:
logger.debug(
'No validate nor apply functions found for {0}'.format(
component))
result[component] = False
return result
def apply(license_string):
"""
Applies a license. It will apply as much licenses as possible, however, it won't fail on invalid licenses as it
will simply skip them.
"""
try:
clients = {}
storagerouters = StorageRouterList.get_storagerouters()
try:
for storagerouter in storagerouters:
clients[storagerouter] = SSHClient(storagerouter.ip)
except UnableToConnectException:
raise RuntimeError('Not all StorageRouters are reachable')
data = LicenseController._decode(license_string)
for component in data:
cdata = data[component]
name = cdata['name']
data = cdata['data']
token = cdata['token']
valid_until = float(cdata['valid_until']) if 'valid_until' in cdata else None
if valid_until is not None and valid_until <= time.time():
continue
signature = cdata['signature'] if 'signature' in cdata else None
validate_functions = Toolbox.fetch_hooks('license', '{0}.validate'.format(component))
apply_functions = Toolbox.fetch_hooks('license', '{0}.apply'.format(component))
if len(validate_functions) == 1 and len(apply_functions) == 1:
valid, metadata = validate_functions[0](component=component, data=data, signature=signature)
if valid is True:
success = apply_functions[0](component=component, data=data, signature=signature)
if success is True:
license_object = LicenseList.get_by_component(component)
if license_object is None:
license_object = License()
license_object.component = component
license_object.name = name
license_object.token = token
license_object.data = data
license_object.valid_until = valid_until
license_object.signature = signature
license_object.save()
license_contents = []
for lic in LicenseList.get_licenses():
license_contents.append(lic.hash)
for storagerouter in storagerouters:
client = clients[storagerouter]
client.file_write('/opt/OpenvStorage/config/licenses', '{0}\n'.format('\n'.join(license_contents)))
except Exception, ex:
LicenseController._logger.exception('Error applying license: {0}'.format(ex))
return None
def refresh_package_information():
"""
Retrieve and store the package information of all StorageRouters
:return: None
"""
GenericController._logger.info('Updating package information')
threads = []
information = {}
all_storagerouters = StorageRouterList.get_storagerouters()
for storagerouter in all_storagerouters:
information[storagerouter.ip] = {}
for function in Toolbox.fetch_hooks('update', 'get_package_info_multi'):
try:
# We make use of these clients in Threads --> cached = False
client = SSHClient(endpoint=storagerouter, username='******', cached=False)
except UnableToConnectException:
information[storagerouter.ip]['errors'] = ['StorageRouter {0} is inaccessible'.format(storagerouter.name)]
break
thread = Thread(target=function,
args=(client, information))
thread.start()
threads.append(thread)
for function in Toolbox.fetch_hooks('update', 'get_package_info_single'):
thread = Thread(target=function,
args=(information,))
thread.start()
threads.append(thread)
for thread in threads:
thread.join()
errors = []
copy_information = copy.deepcopy(information)
for ip, info in information.iteritems():
if len(info.get('errors', [])) > 0:
errors.extend(['{0}: {1}'.format(ip, error) for error in info['errors']])
copy_information.pop(ip)
for storagerouter in all_storagerouters:
info = copy_information.get(storagerouter.ip, {})
if 'errors' in info:
info.pop('errors')
storagerouter.package_information = info
storagerouter.save()
if len(errors) > 0:
errors = [str(error) for error in set(errors)]
raise Exception(' - {0}'.format('\n - '.join(errors)))
def remove(license_guid):
"""
Removes a license
"""
clients = {}
storagerouters = StorageRouterList.get_storagerouters()
try:
for storagerouter in storagerouters:
clients[storagerouter] = SSHClient(storagerouter.ip)
except UnableToConnectException:
raise RuntimeError('Not all StorageRouters are reachable')
lic = License(license_guid)
if lic.can_remove is True:
remove_functions = Toolbox.fetch_hooks('license', '{0}.remove'.format(lic.component))
result = remove_functions[0](component=lic.component, data=lic.data, valid_until=lic.valid_until, signature=lic.signature)
if result is True:
lic.delete()
license_contents = []
for lic in LicenseList.get_licenses():
license_contents.append(lic.hash)
for storagerouter in storagerouters:
client = clients[storagerouter]
client.file_write('/opt/OpenvStorage/config/licenses', '{0}\n'.format('\n'.join(license_contents)))
return result
return None
def remove(license_guid):
"""
Removes a license
"""
clients = {}
storagerouters = StorageRouterList.get_storagerouters()
try:
for storagerouter in storagerouters:
clients[storagerouter] = SSHClient(storagerouter.ip)
except UnableToConnectException:
raise RuntimeError('Not all StorageRouters are reachable')
lic = License(license_guid)
if lic.can_remove is True:
remove_functions = Toolbox.fetch_hooks(
'license', '{0}.remove'.format(lic.component))
result = remove_functions[0](component=lic.component,
data=lic.data,
valid_until=lic.valid_until,
signature=lic.signature)
if result is True:
lic.delete()
license_contents = []
for lic in LicenseList.get_licenses():
license_contents.append(lic.hash)
for storagerouter in storagerouters:
client = clients[storagerouter]
client.file_write(
'/opt/OpenvStorage/config/licenses',
'{0}\n'.format('\n'.join(license_contents)))
return result
return None
def _get_package_information(self):
versions_dict = {self._client.ip: {}}
# ALba is always installed with OpenvStorage. The current split however offloads retrieving Alba information to the AlbaNode which is not
# present for non openvstorage-hc installs. Therefore explicititely request the alba information like this (otherwise update will get jeopardized).
final_dict = {}
threads = []
for fct in Toolbox.fetch_hooks(
component='update', sub_component='get_package_info_cluster'):
thread = Thread(target=fct, args=(self._client, versions_dict))
thread.start()
threads.append(thread)
for thread in threads:
thread.join()
for versions in versions_dict[self._client.ip].itervalues():
for package, version in versions.iteritems():
if package in final_dict:
if version != final_dict[package]:
final_dict[package] = min(version, final_dict[package])
else:
final_dict[package] = version
return OrderedDict(
(key, self._stringify_looseversion(value))
for key, value in sorted(final_dict.items(), key=lambda v: v[0]))
def _can_remove(self):
"""
Can be removed
"""
return len(
Toolbox.fetch_hooks('license',
'{0}.remove'.format(self.component))) == 1
def run_backend_domain_hooks(backend_guid):
"""
Run hooks when the Backend Domains have been updated
:param backend_guid: Guid of the Backend to update
:type backend_guid: str
:return: None
"""
for function in Toolbox.fetch_hooks('backend', 'domains-update'):
function(backend_guid=backend_guid)
def run_backend_domain_hooks(backend_guid):
"""
Run hooks when the Backend Domains have been updated
:param backend_guid: Guid of the Backend to update
:type backend_guid: str
:return: None
"""
for fct in Toolbox.fetch_hooks('backend', 'domains-update'):
fct(backend_guid=backend_guid)
def install_plugins():
"""
(Re)load plugins
"""
if ServiceManager.has_service('ovs-watcher-framework',
SSHClient('127.0.0.1', username='******')):
# If the watcher is running, 'ovs setup' was executed and we need to restart everything to load
# the plugin. In the other case, the plugin will be loaded once 'ovs setup' is executed
from ovs.dal.lists.storagerouterlist import StorageRouterList
clients = []
try:
for storagerouter in StorageRouterList.get_storagerouters():
clients.append(SSHClient(storagerouter, username='******'))
except UnableToConnectException:
raise RuntimeError('Not all StorageRouters are reachable')
for client in clients:
for service_name in ['watcher-framework', 'memcached']:
ServiceManager.stop_service(service_name, client=client)
wait = 30
while wait > 0:
if ServiceManager.get_service_status(
service_name, client=client) is False:
break
time.sleep(1)
wait -= 1
if wait == 0:
raise RuntimeError(
'Could not stop service: {0}'.format(service_name))
for client in clients:
for service_name in ['memcached', 'watcher-framework']:
ServiceManager.start_service(service_name, client=client)
wait = 30
while wait > 0:
if ServiceManager.get_service_status(
service_name, client=client) is True:
break
time.sleep(1)
wait -= 1
if wait == 0:
raise RuntimeError(
'Could not start service: {0}'.format(
service_name))
from ovs.dal.helpers import Migration
Migration.migrate()
from ovs.lib.helpers.toolbox import Toolbox
ip = System.get_my_storagerouter().ip
functions = Toolbox.fetch_hooks('plugin', 'postinstall')
for function in functions:
function(ip=ip)
def validate(license_string):
"""
Validates a license with the various components
"""
try:
result = {}
data = LicenseController._decode(license_string)
for component in data:
cdata = data[component]
name = cdata['name']
data = cdata['data']
_ = cdata['token']
valid_until = float(cdata['valid_until']) if 'valid_until' in cdata else None
if valid_until is not None and valid_until <= time.time():
result[component] = False
continue
signature = cdata['signature'] if 'signature' in cdata else None
validate_functions = Toolbox.fetch_hooks('license', '{0}.validate'.format(component))
apply_functions = Toolbox.fetch_hooks('license', '{0}.apply'.format(component))
if len(validate_functions) == 1 and len(apply_functions) == 1:
try:
valid, metadata = validate_functions[0](component=component, data=data, signature=signature)
except Exception, ex:
LicenseController._logger.debug('Error validating license for {0}: {1}'.format(component, ex))
valid = False
metadata = None
if valid is False:
LicenseController._logger.debug('Invalid license for {0}: {1}'.format(component, license_string))
result[component] = False
else:
result[component] = {'valid_until': valid_until,
'metadata': metadata,
'name': name}
else:
LicenseController._logger.debug('No validate nor apply functions found for {0}'.format(component))
result[component] = False
return result
def merge_package_information():
"""
Retrieve the package information from the model for both StorageRouters and ALBA Nodes and merge it
:return: Package information for all StorageRouters and ALBA nodes
:rtype: dict
"""
package_info = dict((storagerouter.ip, storagerouter.package_information) for storagerouter in StorageRouterList.get_storagerouters())
for function in Toolbox.fetch_hooks('update', 'merge_package_info'):
output = function()
for ip in output:
if ip in package_info:
package_info[ip].update(output[ip])
else:
package_info[ip] = output[ip]
return package_info
def install_plugins():
"""
(Re)load plugins
"""
manager = ServiceFactory.get_manager()
if manager.has_service('ovs-watcher-framework',
SSHClient('127.0.0.1', username='******')):
# If the watcher is running, 'ovs setup' was executed and we need to restart everything to load
# the plugin. In the other case, the plugin will be loaded once 'ovs setup' is executed
print 'Installing plugin into Open vStorage'
from ovs.dal.lists.storagerouterlist import StorageRouterList
clients = {}
masters = StorageRouterList.get_masters()
slaves = StorageRouterList.get_slaves()
try:
for sr in masters + slaves:
clients[sr] = SSHClient(sr, username='******')
except UnableToConnectException:
raise RuntimeError('Not all StorageRouters are reachable')
memcached = 'memcached'
watcher = 'watcher-framework'
for sr in masters + slaves:
if manager.has_service(watcher, clients[sr]):
print '- Stopping watcher on {0} ({1})'.format(
sr.name, sr.ip)
manager.stop_service(watcher, clients[sr])
for sr in masters:
print '- Restarting memcached on {0} ({1})'.format(
sr.name, sr.ip)
manager.restart_service(memcached, clients[sr])
for sr in masters + slaves:
if manager.has_service(watcher, clients[sr]):
print '- Starting watcher on {0} ({1})'.format(
sr.name, sr.ip)
manager.start_service(watcher, clients[sr])
print '- Execute model migrations'
from ovs.dal.helpers import Migration
Migration.migrate()
from ovs.lib.helpers.toolbox import Toolbox
ip = System.get_my_storagerouter().ip
functions = Toolbox.fetch_hooks('plugin', 'postinstall')
if len(functions) > 0:
print '- Execute post installation scripts'
for fct in functions:
fct(ip=ip)
print 'Installing plugin into Open vStorage: Completed'
def install_plugins():
"""
(Re)load plugins
"""
if ServiceManager.has_service('ovs-watcher-framework', SSHClient('127.0.0.1', username='******')):
# If the watcher is running, 'ovs setup' was executed and we need to restart everything to load
# the plugin. In the other case, the plugin will be loaded once 'ovs setup' is executed
from ovs.dal.lists.storagerouterlist import StorageRouterList
clients = []
try:
for storagerouter in StorageRouterList.get_storagerouters():
clients.append(SSHClient(storagerouter, username='******'))
except UnableToConnectException:
raise RuntimeError('Not all StorageRouters are reachable')
for client in clients:
for service_name in ['watcher-framework', 'memcached']:
ServiceManager.stop_service(service_name, client=client)
wait = 30
while wait > 0:
if ServiceManager.get_service_status(service_name, client=client) is False:
break
time.sleep(1)
wait -= 1
if wait == 0:
raise RuntimeError('Could not stop service: {0}'.format(service_name))
for client in clients:
for service_name in ['memcached', 'watcher-framework']:
ServiceManager.start_service(service_name, client=client)
wait = 30
while wait > 0:
if ServiceManager.get_service_status(service_name, client=client) is True:
break
time.sleep(1)
wait -= 1
if wait == 0:
raise RuntimeError('Could not start service: {0}'.format(service_name))
from ovs.dal.helpers import Migration
Migration.migrate()
from ovs.lib.helpers.toolbox import Toolbox
ip = System.get_my_storagerouter().ip
functions = Toolbox.fetch_hooks('plugin', 'postinstall')
for function in functions:
function(ip=ip)
def install_plugins():
"""
(Re)load plugins
"""
if ServiceManager.has_service('ovs-watcher-framework', SSHClient('127.0.0.1', username='******')):
# If the watcher is running, 'ovs setup' was executed and we need to restart everything to load
# the plugin. In the other case, the plugin will be loaded once 'ovs setup' is executed
print 'Installing plugin into Open vStorage'
from ovs.dal.lists.storagerouterlist import StorageRouterList
clients = {}
masters = StorageRouterList.get_masters()
slaves = StorageRouterList.get_slaves()
try:
for sr in masters + slaves:
clients[sr] = SSHClient(sr, username='******')
except UnableToConnectException:
raise RuntimeError('Not all StorageRouters are reachable')
memcached = 'memcached'
watcher = 'watcher-framework'
for sr in masters + slaves:
if ServiceManager.has_service(watcher, clients[sr]):
print '- Stopping watcher on {0} ({1})'.format(sr.name, sr.ip)
ServiceManager.stop_service(watcher, clients[sr])
for sr in masters:
print '- Restarting memcached on {0} ({1})'.format(sr.name, sr.ip)
ServiceManager.restart_service(memcached, clients[sr])
for sr in masters + slaves:
if ServiceManager.has_service(watcher, clients[sr]):
print '- Starting watcher on {0} ({1})'.format(sr.name, sr.ip)
ServiceManager.start_service(watcher, clients[sr])
print '- Execute model migrations'
from ovs.dal.helpers import Migration
Migration.migrate()
from ovs.lib.helpers.toolbox import Toolbox
ip = System.get_my_storagerouter().ip
functions = Toolbox.fetch_hooks('plugin', 'postinstall')
if len(functions) > 0:
print '- Execute post installation scripts'
for function in functions:
function(ip=ip)
print 'Installing plugin into Open vStorage: Completed'
def get_update_information_all():
"""
Retrieve the update information for all StorageRouters
This contains information about
- downtime of model, GUI, vPools, proxies, ...
- services that will be restarted
- packages that will be updated
- prerequisites that have not been met
:return: Information about the update
:rtype: dict
"""
information = {}
for function in Toolbox.fetch_hooks('update', 'information'):
function(information=information)
for component, info in copy.deepcopy(information).iteritems():
if len(info['packages']) == 0:
information.pop(component)
return information
def update_framework():
"""
Update the framework
:return: None
"""
filemutex = file_mutex('system_update', wait=2)
upgrade_file = '/etc/ready_for_upgrade'
upgrade_ongoing_check_file = '/etc/upgrade_ongoing'
ssh_clients = []
try:
filemutex.acquire()
UpdateController._log_message('+++ Starting framework update +++')
from ovs.dal.lists.storagerouterlist import StorageRouterList
UpdateController._log_message('Generating SSH client connections for each storage router')
upgrade_file = '/etc/ready_for_upgrade'
upgrade_ongoing_check_file = '/etc/upgrade_ongoing'
storage_routers = StorageRouterList.get_storagerouters()
ssh_clients = []
master_ips = []
extra_ips = []
for sr in storage_routers:
ssh_clients.append(SSHClient(sr.ip, username='******'))
if sr.node_type == 'MASTER':
master_ips.append(sr.ip)
elif sr.node_type == 'EXTRA':
extra_ips.append(sr.ip)
this_client = [client for client in ssh_clients if client.is_local is True][0]
# Create locks
UpdateController._log_message('Creating lock files', client_ip=this_client.ip)
for client in ssh_clients:
client.run('touch {0}'.format(upgrade_file)) # Prevents manual install or upgrade individual packages
client.run('touch {0}'.format(upgrade_ongoing_check_file)) # Prevents clicking x times on 'Update' btn
# Check requirements
packages_to_update = set()
all_services_to_restart = []
for client in ssh_clients:
for function in Toolbox.fetch_hooks('update', 'metadata'):
UpdateController._log_message('Executing function {0}'.format(function.__name__),
client_ip=client.ip)
output = function(client)
for key, value in output.iteritems():
if key != 'framework':
continue
for package_info in value:
packages_to_update.update(package_info['packages'])
all_services_to_restart += package_info['services']
services_to_restart = []
for service in all_services_to_restart:
if service not in services_to_restart:
services_to_restart.append(service) # Filter out duplicates maintaining the order of services (eg: watcher-framework before memcached)
UpdateController._log_message('Services which will be restarted --> {0}'.format(', '.join(services_to_restart)))
UpdateController._log_message('Packages which will be installed --> {0}'.format(', '.join(packages_to_update)))
# Stop services
if UpdateController._change_services_state(services=services_to_restart,
ssh_clients=ssh_clients,
action='stop') is False:
UpdateController._log_message('Stopping all services on every node failed, cannot continue',
client_ip=this_client.ip, severity='warning')
UpdateController._remove_lock_files([upgrade_file, upgrade_ongoing_check_file], ssh_clients)
# Start services again if a service could not be stopped
UpdateController._log_message('Attempting to start the services again', client_ip=this_client.ip)
UpdateController._change_services_state(services=services_to_restart,
ssh_clients=ssh_clients,
action='start')
UpdateController._log_message('Failed to stop all required services, aborting update',
client_ip=this_client.ip, severity='error')
return
# Update packages
failed_clients = []
for client in ssh_clients:
PackageManager.update(client=client)
try:
UpdateController._log_message('Installing latest packages', client.ip)
for package in packages_to_update:
UpdateController._log_message('Installing {0}'.format(package), client.ip)
PackageManager.install(package_name=package,
client=client,
force=True)
UpdateController._log_message('Installed {0}'.format(package), client.ip)
client.file_delete(upgrade_file)
except subprocess.CalledProcessError as cpe:
UpdateController._log_message('Upgrade failed with error: {0}'.format(cpe.output), client.ip,
'error')
failed_clients.append(client)
break
if failed_clients:
UpdateController._remove_lock_files([upgrade_file, upgrade_ongoing_check_file], ssh_clients)
UpdateController._log_message('Error occurred. Attempting to start all services again',
client_ip=this_client.ip, severity='error')
UpdateController._change_services_state(services=services_to_restart,
ssh_clients=ssh_clients,
action='start')
UpdateController._log_message('Failed to upgrade following nodes:\n - {0}\nPlease check /var/log/ovs/lib.log on {1} for more information'.format('\n - '.join([client.ip for client in failed_clients]), this_client.ip),
this_client.ip,
'error')
return
# Migrate code
for client in ssh_clients:
try:
UpdateController._log_message('Started code migration', client.ip)
try:
with remote(client.ip, [Migrator]) as rem:
rem.Migrator.migrate(master_ips, extra_ips)
except EOFError as eof:
UpdateController._log_message('EOFError during code migration, retrying {0}'.format(eof), client.ip, 'warning')
with remote(client.ip, [Migrator]) as rem:
rem.Migrator.migrate(master_ips, extra_ips)
UpdateController._log_message('Finished code migration', client.ip)
except Exception as ex:
UpdateController._remove_lock_files([upgrade_ongoing_check_file], ssh_clients)
UpdateController._log_message('Code migration failed with error: {0}'.format(ex), client.ip, 'error')
return
# Start services
UpdateController._log_message('Starting services', client_ip=this_client.ip)
model_services = []
if 'arakoon-ovsdb' in services_to_restart:
model_services.append('arakoon-ovsdb')
services_to_restart.remove('arakoon-ovsdb')
if 'memcached' in services_to_restart:
model_services.append('memcached')
services_to_restart.remove('memcached')
UpdateController._change_services_state(services=model_services,
ssh_clients=ssh_clients,
action='start')
# Migrate model
UpdateController._log_message('Started model migration', client_ip=this_client.ip)
try:
from ovs.dal.helpers import Migration
with remote(ssh_clients[0].ip, [Migration]) as rem:
rem.Migration.migrate()
UpdateController._log_message('Finished model migration', client_ip=this_client.ip)
except Exception as ex:
UpdateController._remove_lock_files([upgrade_ongoing_check_file], ssh_clients)
UpdateController._log_message('An unexpected error occurred: {0}'.format(ex), client_ip=this_client.ip,
severity='error')
return
# Post upgrade actions
UpdateController._log_message('Executing post upgrade actions', client_ip=this_client.ip)
for client in ssh_clients:
with remote(client.ip, [Toolbox, SSHClient]) as rem:
for function in rem.Toolbox.fetch_hooks('update', 'postupgrade'):
UpdateController._log_message('Executing action {0}'.format(function.__name__),
client_ip=client.ip)
try:
function(rem.SSHClient(client.ip, username='******'))
UpdateController._log_message('Executing action {0} completed'.format(function.__name__),
client_ip=client.ip)
except Exception as ex:
UpdateController._log_message('Post upgrade action failed with error: {0}'.format(ex),
client.ip, 'error')
# Start watcher and restart support-agent
UpdateController._change_services_state(services=services_to_restart,
ssh_clients=ssh_clients,
action='start')
UpdateController._change_services_state(services=['support-agent'],
ssh_clients=ssh_clients,
action='restart')
UpdateController._remove_lock_files([upgrade_ongoing_check_file], ssh_clients)
UpdateController._log_message('+++ Finished updating +++')
except RuntimeError as rte:
UpdateController._log_message('Error during framework update: {0}'.format(rte), severity='error')
UpdateController._remove_lock_files([upgrade_file, upgrade_ongoing_check_file], ssh_clients)
except NoLockAvailableException:
UpdateController._log_message('Another framework update is currently in progress!')
except Exception as ex:
UpdateController._log_message('Error during framework update: {0}'.format(ex), severity='error')
UpdateController._remove_lock_files([upgrade_file, upgrade_ongoing_check_file], ssh_clients)
finally:
filemutex.release()
def remove_node(node_ip, silent=None):
"""
Remove the node with specified IP from the cluster
:param node_ip: IP of the node to remove
:type node_ip: str
:param silent: If silent == '--force-yes' no question will be asked to confirm the removal
:type silent: str
:return: None
"""
from ovs.dal.lists.storagerouterlist import StorageRouterList
from ovs.lib.storagedriver import StorageDriverController
from ovs.lib.vpool import VPoolController
Toolbox.log(logger=NodeRemovalController._logger,
messages='Remove node',
boxed=True)
Toolbox.log(
logger=NodeRemovalController._logger,
messages=
'WARNING: Some of these steps may take a very long time, please check the logs for more information\n\n'
)
service_manager = ServiceFactory.get_manager()
###############
# VALIDATIONS #
###############
try:
node_ip = node_ip.strip()
if not isinstance(node_ip, str):
raise ValueError('Node IP must be a string')
if not re.match(SSHClient.IP_REGEX, node_ip):
raise ValueError('Invalid IP {0} specified'.format(node_ip))
storage_router_all = sorted(StorageRouterList.get_storagerouters(),
key=lambda k: k.name)
storage_router_masters = StorageRouterList.get_masters()
storage_router_all_ips = set(
[storage_router.ip for storage_router in storage_router_all])
storage_router_master_ips = set([
storage_router.ip for storage_router in storage_router_masters
])
storage_router_to_remove = StorageRouterList.get_by_ip(node_ip)
offline_reasons = {}
if node_ip not in storage_router_all_ips:
raise ValueError(
'Unknown IP specified\nKnown in model:\n - {0}\nSpecified for removal:\n - {1}'
.format('\n - '.join(storage_router_all_ips), node_ip))
if len(storage_router_all_ips) == 1:
raise RuntimeError("Removing the only node is not possible")
if node_ip in storage_router_master_ips and len(
storage_router_master_ips) == 1:
raise RuntimeError(
"Removing the only master node is not possible")
if System.get_my_storagerouter() == storage_router_to_remove:
raise RuntimeError(
'The node to be removed cannot be identical to the node on which the removal is initiated'
)
Toolbox.log(
logger=NodeRemovalController._logger,
messages='Creating SSH connections to remaining master nodes')
master_ip = None
ip_client_map = {}
storage_routers_offline = []
storage_router_to_remove_online = True
for storage_router in storage_router_all:
try:
client = SSHClient(storage_router,
username='******',
timeout=10)
except (UnableToConnectException, NotAuthenticatedException,
TimeOutException) as ex:
if isinstance(ex, UnableToConnectException):
msg = 'Unable to connect'
elif isinstance(ex, NotAuthenticatedException):
msg = 'Could not authenticate'
elif isinstance(ex, TimeOutException):
msg = 'Connection timed out'
Toolbox.log(
logger=NodeRemovalController._logger,
messages=' * Node with IP {0:<15}- {1}'.format(
storage_router.ip, msg))
offline_reasons[storage_router.ip] = msg
storage_routers_offline.append(storage_router)
if storage_router == storage_router_to_remove:
storage_router_to_remove_online = False
continue
Toolbox.log(
logger=NodeRemovalController._logger,
messages=' * Node with IP {0:<15}- Successfully connected'
.format(storage_router.ip))
ip_client_map[storage_router.ip] = client
if storage_router != storage_router_to_remove and storage_router.node_type == 'MASTER':
master_ip = storage_router.ip
if len(ip_client_map) == 0 or master_ip is None:
raise RuntimeError(
'Could not connect to any master node in the cluster')
storage_router_to_remove.invalidate_dynamics('vdisks_guids')
if len(
storage_router_to_remove.vdisks_guids
) > 0: # vDisks are supposed to be moved away manually before removing a node
raise RuntimeError(
"Still vDisks attached to Storage Router {0}".format(
storage_router_to_remove.name))
internal_memcached = Toolbox.is_service_internally_managed(
service='memcached')
internal_rabbit_mq = Toolbox.is_service_internally_managed(
service='rabbitmq')
memcached_endpoints = Configuration.get(
key='/ovs/framework/memcache|endpoints')
rabbit_mq_endpoints = Configuration.get(
key='/ovs/framework/messagequeue|endpoints')
copy_memcached_endpoints = list(memcached_endpoints)
copy_rabbit_mq_endpoints = list(rabbit_mq_endpoints)
for endpoint in memcached_endpoints:
if endpoint.startswith(storage_router_to_remove.ip):
copy_memcached_endpoints.remove(endpoint)
for endpoint in rabbit_mq_endpoints:
if endpoint.startswith(storage_router_to_remove.ip):
copy_rabbit_mq_endpoints.remove(endpoint)
if len(copy_memcached_endpoints
) == 0 and internal_memcached is True:
raise RuntimeError(
'Removal of provided nodes will result in a complete removal of the memcached service'
)
if len(copy_rabbit_mq_endpoints
) == 0 and internal_rabbit_mq is True:
raise RuntimeError(
'Removal of provided nodes will result in a complete removal of the messagequeue service'
)
Toolbox.run_hooks(component='noderemoval',
sub_component='validate_removal',
logger=NodeRemovalController._logger,
cluster_ip=storage_router_to_remove.ip)
except KeyboardInterrupt:
Toolbox.log(logger=NodeRemovalController._logger, messages='\n')
Toolbox.log(
logger=NodeRemovalController._logger,
messages=
'Removal has been aborted during the validation step. No changes have been applied.',
boxed=True,
loglevel='warning')
sys.exit(1)
except Exception as exception:
Toolbox.log(logger=NodeRemovalController._logger,
messages=[str(exception)],
boxed=True,
loglevel='exception')
sys.exit(1)
#################
# CONFIRMATIONS #
#################
try:
interactive = silent != '--force-yes'
remove_asd_manager = not interactive # Remove ASD manager if non-interactive else ask
if interactive is True:
if len(storage_routers_offline) > 0:
Toolbox.log(
logger=NodeRemovalController._logger,
messages=
'Certain nodes appear to be offline. These will not fully removed and will cause issues if they are not really offline.'
)
Toolbox.log(
logger=NodeRemovalController._logger,
messages='Offline nodes: {0}'.format(''.join(
('\n * {0:<15}- {1}.'.format(ip, message)
for ip, message in offline_reasons.iteritems()))))
valid_node_info = Interactive.ask_yesno(
message=
'Continue the removal with these being presumably offline?',
default_value=False)
if valid_node_info is False:
Toolbox.log(
logger=NodeRemovalController._logger,
messages=
'Please validate the state of the nodes before removing.',
title=True)
sys.exit(1)
proceed = Interactive.ask_yesno(
message='Are you sure you want to remove node {0}?'.format(
storage_router_to_remove.name),
default_value=False)
if proceed is False:
Toolbox.log(logger=NodeRemovalController._logger,
messages='Abort removal',
title=True)
sys.exit(1)
remove_asd_manager = True
if storage_router_to_remove_online is True:
client = SSHClient(endpoint=storage_router_to_remove,
username='******')
if service_manager.has_service(name='asd-manager',
client=client):
remove_asd_manager = Interactive.ask_yesno(
message=
'Do you also want to remove the ASD manager and related ASDs?',
default_value=False)
if remove_asd_manager is True or storage_router_to_remove_online is False:
for fct in Toolbox.fetch_hooks('noderemoval',
'validate_asd_removal'):
validation_output = fct(storage_router_to_remove.ip)
if validation_output['confirm'] is True:
if Interactive.ask_yesno(
message=validation_output['question'],
default_value=False) is False:
remove_asd_manager = False
break
except KeyboardInterrupt:
Toolbox.log(logger=NodeRemovalController._logger, messages='\n')
Toolbox.log(
logger=NodeRemovalController._logger,
messages=
'Removal has been aborted during the confirmation step. No changes have been applied.',
boxed=True,
loglevel='warning')
sys.exit(1)
except Exception as exception:
Toolbox.log(logger=NodeRemovalController._logger,
messages=[str(exception)],
boxed=True,
loglevel='exception')
sys.exit(1)
###########
# REMOVAL #
###########
try:
Toolbox.log(logger=NodeRemovalController._logger,
messages='Starting removal of node {0} - {1}'.format(
storage_router_to_remove.name,
storage_router_to_remove.ip))
if storage_router_to_remove_online is False:
Toolbox.log(
logger=NodeRemovalController._logger,
messages=
' Marking all Storage Drivers served by Storage Router {0} as offline'
.format(storage_router_to_remove.ip))
StorageDriverController.mark_offline(
storagerouter_guid=storage_router_to_remove.guid)
# Remove vPools
Toolbox.log(logger=NodeRemovalController._logger,
messages=' Removing vPools from node'.format(
storage_router_to_remove.ip))
storage_routers_offline_guids = [
sr.guid for sr in storage_routers_offline
if sr.guid != storage_router_to_remove.guid
]
for storage_driver in storage_router_to_remove.storagedrivers:
Toolbox.log(logger=NodeRemovalController._logger,
messages=' Removing vPool {0} from node'.format(
storage_driver.vpool.name))
VPoolController.shrink_vpool(
storagedriver_guid=storage_driver.guid,
offline_storage_router_guids=storage_routers_offline_guids)
# Demote if MASTER
if storage_router_to_remove.node_type == 'MASTER':
NodeTypeController.demote_node(
cluster_ip=storage_router_to_remove.ip,
master_ip=master_ip,
ip_client_map=ip_client_map,
unique_id=storage_router_to_remove.machine_id,
unconfigure_memcached=internal_memcached,
unconfigure_rabbitmq=internal_rabbit_mq,
offline_nodes=storage_routers_offline)
# Stop / remove services
Toolbox.log(logger=NodeRemovalController._logger,
messages='Stopping and removing services')
if storage_router_to_remove_online is True:
client = SSHClient(endpoint=storage_router_to_remove,
username='******')
NodeRemovalController.remove_services(
client=client,
node_type=storage_router_to_remove.node_type.lower(),
logger=NodeRemovalController._logger)
service = 'watcher-config'
if service_manager.has_service(service, client=client):
Toolbox.log(
logger=NodeRemovalController._logger,
messages='Removing service {0}'.format(service))
service_manager.stop_service(service, client=client)
service_manager.remove_service(service, client=client)
Toolbox.run_hooks(component='noderemoval',
sub_component='remove',
logger=NodeRemovalController._logger,
cluster_ip=storage_router_to_remove.ip,
complete_removal=remove_asd_manager)
# Clean up model
Toolbox.log(logger=NodeRemovalController._logger,
messages='Removing node from model')
for service in storage_router_to_remove.services:
service.delete()
for disk in storage_router_to_remove.disks:
for partition in disk.partitions:
partition.delete()
disk.delete()
for j_domain in storage_router_to_remove.domains:
j_domain.delete()
Configuration.delete('/ovs/framework/hosts/{0}'.format(
storage_router_to_remove.machine_id))
NodeTypeController.restart_framework_and_memcache_services(
clients=ip_client_map,
offline_node_ips=[node.ip for node in storage_routers_offline],
logger=NodeRemovalController._logger)
if storage_router_to_remove_online is True:
client = SSHClient(endpoint=storage_router_to_remove,
username='******')
client.file_delete(filenames=[CACC_LOCATION])
client.file_delete(filenames=[CONFIG_STORE_LOCATION])
storage_router_to_remove.delete()
Toolbox.log(logger=NodeRemovalController._logger,
messages='Successfully removed node\n')
except Exception as exception:
Toolbox.log(logger=NodeRemovalController._logger, messages='\n')
Toolbox.log(
logger=NodeRemovalController._logger,
messages=['An unexpected error occurred:',
str(exception)],
boxed=True,
loglevel='exception')
sys.exit(1)
except KeyboardInterrupt:
Toolbox.log(logger=NodeRemovalController._logger, messages='\n')
Toolbox.log(
logger=NodeRemovalController._logger,
messages=
'This setup was aborted. Open vStorage may be in an inconsistent state, make sure to validate the installation.',
boxed=True,
loglevel='error')
sys.exit(1)
if remove_asd_manager is True and storage_router_to_remove_online is True:
Toolbox.log(logger=NodeRemovalController._logger,
messages='\nRemoving ASD Manager')
with remote(storage_router_to_remove.ip, [os]) as rem:
rem.os.system('asd-manager remove --force-yes')
Toolbox.log(logger=NodeRemovalController._logger,
messages='Remove nodes finished',
title=True)
def update_framework():
"""
Update the framework
:return: None
"""
file_mutex = FileMutex('system_update', wait=2)
upgrade_file = '/etc/ready_for_upgrade'
upgrade_ongoing_check_file = '/etc/upgrade_ongoing'
ssh_clients = []
try:
file_mutex.acquire()
UpdateController._log_message('+++ Starting framework update +++')
from ovs.dal.lists.storagerouterlist import StorageRouterList
UpdateController._log_message(
'Generating SSH client connections for each storage router')
upgrade_file = '/etc/ready_for_upgrade'
upgrade_ongoing_check_file = '/etc/upgrade_ongoing'
storage_routers = StorageRouterList.get_storagerouters()
ssh_clients = []
master_ips = []
extra_ips = []
for sr in storage_routers:
ssh_clients.append(SSHClient(sr.ip, username='******'))
if sr.node_type == 'MASTER':
master_ips.append(sr.ip)
elif sr.node_type == 'EXTRA':
extra_ips.append(sr.ip)
this_client = [
client for client in ssh_clients if client.is_local is True
][0]
# Create locks
UpdateController._log_message('Creating lock files',
client_ip=this_client.ip)
for client in ssh_clients:
client.run(
'touch {0}'.format(upgrade_file)
) # Prevents manual install or upgrade individual packages
client.run('touch {0}'.format(upgrade_ongoing_check_file)
) # Prevents clicking x times on 'Update' btn
# Check requirements
packages_to_update = set()
all_services_to_restart = []
for client in ssh_clients:
for function in Toolbox.fetch_hooks('update', 'metadata'):
UpdateController._log_message(
'Executing function {0}'.format(function.__name__),
client_ip=client.ip)
output = function(client)
for key, value in output.iteritems():
if key != 'framework':
continue
for package_info in value:
packages_to_update.update(package_info['packages'])
all_services_to_restart += package_info['services']
services_to_restart = []
for service in all_services_to_restart:
if service not in services_to_restart:
services_to_restart.append(
service
) # Filter out duplicates maintaining the order of services (eg: watcher-framework before memcached)
UpdateController._log_message(
'Services which will be restarted --> {0}'.format(
', '.join(services_to_restart)))
UpdateController._log_message(
'Packages which will be installed --> {0}'.format(
', '.join(packages_to_update)))
# Stop services
if UpdateController._change_services_state(
services=services_to_restart,
ssh_clients=ssh_clients,
action='stop') is False:
UpdateController._log_message(
'Stopping all services on every node failed, cannot continue',
client_ip=this_client.ip,
severity='warning')
UpdateController._remove_lock_files(
[upgrade_file, upgrade_ongoing_check_file], ssh_clients)
# Start services again if a service could not be stopped
UpdateController._log_message(
'Attempting to start the services again',
client_ip=this_client.ip)
UpdateController._change_services_state(
services=services_to_restart,
ssh_clients=ssh_clients,
action='start')
UpdateController._log_message(
'Failed to stop all required services, aborting update',
client_ip=this_client.ip,
severity='error')
return
# Update packages
failed_clients = []
for client in ssh_clients:
PackageManager.update(client=client)
try:
UpdateController._log_message('Installing latest packages',
client.ip)
for package in packages_to_update:
UpdateController._log_message(
'Installing {0}'.format(package), client.ip)
PackageManager.install(package_name=package,
client=client,
force=True)
UpdateController._log_message(
'Installed {0}'.format(package), client.ip)
client.file_delete(upgrade_file)
except subprocess.CalledProcessError as cpe:
UpdateController._log_message(
'Upgrade failed with error: {0}'.format(cpe.output),
client.ip, 'error')
failed_clients.append(client)
break
if failed_clients:
UpdateController._remove_lock_files(
[upgrade_file, upgrade_ongoing_check_file], ssh_clients)
UpdateController._log_message(
'Error occurred. Attempting to start all services again',
client_ip=this_client.ip,
severity='error')
UpdateController._change_services_state(
services=services_to_restart,
ssh_clients=ssh_clients,
action='start')
UpdateController._log_message(
'Failed to upgrade following nodes:\n - {0}\nPlease check /var/log/ovs/lib.log on {1} for more information'
.format('\n - '.join([
client.ip for client in failed_clients
])), this_client.ip, 'error')
return
# Migrate code
for client in ssh_clients:
try:
UpdateController._log_message('Started code migration',
client.ip)
try:
with Remote(client.ip, [Migrator]) as remote:
remote.Migrator.migrate(master_ips, extra_ips)
except EOFError as eof:
UpdateController._log_message(
'EOFError during code migration, retrying {0}'.
format(eof), client.ip, 'warning')
with Remote(client.ip, [Migrator]) as remote:
remote.Migrator.migrate(master_ips, extra_ips)
UpdateController._log_message('Finished code migration',
client.ip)
except Exception as ex:
UpdateController._remove_lock_files(
[upgrade_ongoing_check_file], ssh_clients)
UpdateController._log_message(
'Code migration failed with error: {0}'.format(ex),
client.ip, 'error')
return
# Start services
UpdateController._log_message('Starting services',
client_ip=this_client.ip)
model_services = []
if 'arakoon-ovsdb' in services_to_restart:
model_services.append('arakoon-ovsdb')
services_to_restart.remove('arakoon-ovsdb')
if 'memcached' in services_to_restart:
model_services.append('memcached')
services_to_restart.remove('memcached')
UpdateController._change_services_state(services=model_services,
ssh_clients=ssh_clients,
action='start')
# Migrate model
UpdateController._log_message('Started model migration',
client_ip=this_client.ip)
try:
from ovs.dal.helpers import Migration
Migration.migrate()
UpdateController._log_message('Finished model migration',
client_ip=this_client.ip)
except Exception as ex:
UpdateController._remove_lock_files(
[upgrade_ongoing_check_file], ssh_clients)
UpdateController._log_message(
'An unexpected error occurred: {0}'.format(ex),
client_ip=this_client.ip,
severity='error')
return
# Post upgrade actions
UpdateController._log_message('Executing post upgrade actions',
client_ip=this_client.ip)
for client in ssh_clients:
with Remote(client.ip, [Toolbox, SSHClient]) as remote:
for function in remote.Toolbox.fetch_hooks(
'update', 'postupgrade'):
UpdateController._log_message(
'Executing action {0}'.format(function.__name__),
client_ip=client.ip)
try:
function(
remote.SSHClient(client.ip, username='******'))
UpdateController._log_message(
'Executing action {0} completed'.format(
function.__name__),
client_ip=client.ip)
except Exception as ex:
UpdateController._log_message(
'Post upgrade action failed with error: {0}'.
format(ex), client.ip, 'error')
# Start watcher and restart support-agent
UpdateController._change_services_state(
services=services_to_restart,
ssh_clients=ssh_clients,
action='start')
UpdateController._change_services_state(services=['support-agent'],
ssh_clients=ssh_clients,
action='restart')
UpdateController._remove_lock_files([upgrade_ongoing_check_file],
ssh_clients)
UpdateController._log_message('+++ Finished updating +++')
except RuntimeError as rte:
if 'Could not acquire lock' in rte.message:
UpdateController._log_message(
'Another framework update is currently in progress!')
else:
UpdateController._log_message(
'Error during framework update: {0}'.format(rte),
severity='error')
UpdateController._remove_lock_files(
[upgrade_file, upgrade_ongoing_check_file], ssh_clients)
except Exception as ex:
UpdateController._log_message(
'Error during framework update: {0}'.format(ex),
severity='error')
UpdateController._remove_lock_files(
[upgrade_file, upgrade_ongoing_check_file], ssh_clients)
finally:
file_mutex.release()
def update_volumedriver():
"""
Update the volumedriver
:return: None
"""
file_mutex = FileMutex('system_update', wait=2)
upgrade_file = '/etc/ready_for_upgrade'
upgrade_ongoing_check_file = '/etc/upgrade_ongoing'
ssh_clients = []
try:
file_mutex.acquire()
UpdateController._log_message(
'+++ Starting volumedriver update +++')
from ovs.dal.lists.storagerouterlist import StorageRouterList
UpdateController._log_message(
'Generating SSH client connections for each storage router')
storage_routers = StorageRouterList.get_storagerouters()
ssh_clients = [
SSHClient(storage_router.ip, 'root')
for storage_router in storage_routers
]
this_client = [
client for client in ssh_clients if client.is_local is True
][0]
# Commence update !!!!!!!
# 0. Create locks
UpdateController._log_message('Creating lock files',
client_ip=this_client.ip)
for client in ssh_clients:
client.run(
'touch {0}'.format(upgrade_file)
) # Prevents manual install or upgrade individual packages
client.run('touch {0}'.format(upgrade_ongoing_check_file)
) # Prevents clicking x times on 'Update' btn
# 1. Check requirements
packages_to_update = set()
all_services_to_restart = []
for client in ssh_clients:
for function in Toolbox.fetch_hooks('update', 'metadata'):
UpdateController._log_message(
'Executing function {0}'.format(function.__name__),
client_ip=client.ip)
output = function(client)
for key, value in output.iteritems():
if key != 'volumedriver':
continue
for package_info in value:
packages_to_update.update(package_info['packages'])
all_services_to_restart += package_info['services']
services_to_restart = []
for service in all_services_to_restart:
if service not in services_to_restart:
services_to_restart.append(
service
) # Filter out duplicates keeping the order of services (eg: watcher-framework before memcached)
UpdateController._log_message(
'Services which will be restarted --> {0}'.format(
', '.join(services_to_restart)))
UpdateController._log_message(
'Packages which will be installed --> {0}'.format(
', '.join(packages_to_update)))
# 1. Stop services
if UpdateController._change_services_state(
services=services_to_restart,
ssh_clients=ssh_clients,
action='stop') is False:
UpdateController._log_message(
'Stopping all services on every node failed, cannot continue',
client_ip=this_client.ip,
severity='warning')
UpdateController._remove_lock_files(
[upgrade_file, upgrade_ongoing_check_file], ssh_clients)
UpdateController._log_message(
'Attempting to start the services again',
client_ip=this_client.ip)
UpdateController._change_services_state(
services=services_to_restart,
ssh_clients=ssh_clients,
action='start')
UpdateController._log_message(
'Failed to stop all required services, update aborted',
client_ip=this_client.ip,
severity='error')
return
# 2. Update packages
failed_clients = []
for client in ssh_clients:
PackageManager.update(client=client)
try:
for package_name in packages_to_update:
UpdateController._log_message(
'Installing {0}'.format(package_name), client.ip)
PackageManager.install(package_name=package_name,
client=client,
force=True)
UpdateController._log_message(
'Installed {0}'.format(package_name), client.ip)
client.file_delete(upgrade_file)
except subprocess.CalledProcessError as cpe:
UpdateController._log_message(
'Upgrade failed with error: {0}'.format(cpe.output),
client.ip, 'error')
failed_clients.append(client)
break
if failed_clients:
UpdateController._remove_lock_files(
[upgrade_file, upgrade_ongoing_check_file], ssh_clients)
UpdateController._log_message(
'Error occurred. Attempting to start all services again',
client_ip=this_client.ip,
severity='error')
UpdateController._change_services_state(
services=services_to_restart,
ssh_clients=ssh_clients,
action='start')
UpdateController._log_message(
'Failed to upgrade following nodes:\n - {0}\nPlease check /var/log/ovs/lib.log on {1} for more information'
.format('\n - '.join([
client.ip for client in failed_clients
])), this_client.ip, 'error')
return
# 3. Post upgrade actions
UpdateController._log_message('Executing post upgrade actions',
client_ip=this_client.ip)
for client in ssh_clients:
for function in Toolbox.fetch_hooks('update', 'postupgrade'):
UpdateController._log_message(
'Executing action: {0}'.format(function.__name__),
client_ip=client.ip)
try:
function(client)
except Exception as ex:
UpdateController._log_message(
'Post upgrade action failed with error: {0}'.
format(ex), client.ip, 'error')
# 4. Start services
UpdateController._log_message('Starting services',
client_ip=this_client.ip)
UpdateController._change_services_state(
services=services_to_restart,
ssh_clients=ssh_clients,
action='start')
UpdateController._remove_lock_files([upgrade_ongoing_check_file],
ssh_clients)
UpdateController._log_message('+++ Finished updating +++')
except RuntimeError as rte:
if 'Could not acquire lock' in rte.message:
UpdateController._log_message(
'Another volumedriver update is currently in progress!')
else:
UpdateController._log_message(
'Error during volumedriver update: {0}'.format(rte),
severity='error')
UpdateController._remove_lock_files(
[upgrade_file, upgrade_ongoing_check_file], ssh_clients)
except Exception as ex:
UpdateController._log_message(
'Error during volumedriver update: {0}'.format(ex),
severity='error')
UpdateController._remove_lock_files(
[upgrade_file, upgrade_ongoing_check_file], ssh_clients)
finally:
file_mutex.release()
def execute_update(components):
"""
Update the specified components on all StorageRouters
This is called upon by 'at'
:return: None
"""
filemutex = file_mutex('system_update', wait=2)
ssh_clients = []
services_stop_start = set()
try:
filemutex.acquire()
UpdateController._logger.debug('+++ Starting update +++')
from ovs.dal.lists.storagerouterlist import StorageRouterList
# Create SSHClients to all nodes
UpdateController._logger.debug('Generating SSH client connections for each storage router')
storage_routers = StorageRouterList.get_storagerouters()
master_ips = []
extra_ips = []
for sr in storage_routers:
try:
ssh_clients.append(SSHClient(sr.ip, username='******'))
if sr.node_type == 'MASTER':
master_ips.append(sr.ip)
elif sr.node_type == 'EXTRA':
extra_ips.append(sr.ip)
except UnableToConnectException:
raise Exception('Update is only allowed on systems where all nodes are online and fully functional')
# Create locks
for client in ssh_clients:
UpdateController._logger.debug('{0}: Creating lock files'.format(client.ip))
client.run(['touch', UpdateController._update_file]) # Prevents manual install or update individual packages
client.run(['touch', UpdateController._update_ongoing_file])
# Check requirements
packages_to_update = {}
services_post_update = set()
update_information = UpdateController.get_update_information_all()
for component, component_info in update_information.iteritems():
if component in components:
UpdateController._logger.debug('Verifying update information for component: {0}'.format(component.upper()))
Toolbox.verify_required_params(actual_params=component_info,
required_params={'downtime': (list, None),
'packages': (dict, None),
'prerequisites': (list, None),
'services_stop_start': (set, None),
'services_post_update': (set, None)})
if len(component_info['prerequisites']) > 0:
raise Exception('Update is only allowed when all prerequisites have been met')
packages_to_update.update(component_info['packages'])
services_stop_start.update(component_info['services_stop_start'])
services_post_update.update(component_info['services_post_update'])
if len(packages_to_update) > 0:
UpdateController._logger.debug('Packages to be updated: {0}'.format(', '.join(sorted(packages_to_update.keys()))))
if len(services_stop_start) > 0:
UpdateController._logger.debug('Services to stop before package update: {0}'.format(', '.join(sorted(services_stop_start))))
if len(services_post_update) > 0:
UpdateController._logger.debug('Services which will be restarted after update: {0}'.format(', '.join(sorted(services_post_update))))
# Stop services
if UpdateController.change_services_state(services=services_stop_start,
ssh_clients=ssh_clients,
action='stop') is False:
raise Exception('Stopping all services on every node failed, cannot continue')
# Install packages
# First install packages on all StorageRouters individually
if packages_to_update:
failures = False
for client in ssh_clients:
UpdateController._logger.debug('{0}: Installing packages'.format(client.ip))
for function in Toolbox.fetch_hooks('update', 'package_install_multi'):
try:
function(client=client, package_info=packages_to_update, components=components)
except Exception as ex:
UpdateController._logger.error('{0}: Package installation hook {1} failed with error: {2}'.format(client.ip, function.__name__, ex))
failures = True
if set(components).difference({'framework', 'storagedriver'}):
# Second install packages on all ALBA nodes
for function in Toolbox.fetch_hooks('update', 'package_install_single'):
try:
function(package_info=packages_to_update, components=components)
except Exception as ex:
UpdateController._logger.exception('Package installation hook {0} failed with error: {1}'.format(function.__name__, ex))
failures = True
if failures is True:
raise Exception('Installing the packages failed on 1 or more nodes')
# Remove update file
for client in ssh_clients:
client.file_delete(UpdateController._update_file)
# Migrate code
if 'framework' in components:
failures = []
for client in ssh_clients:
UpdateController._logger.debug('{0}: Verifying extensions code migration is required'.format(client.ip))
try:
key = '/ovs/framework/hosts/{0}/versions'.format(System.get_my_machine_id(client=client))
old_versions = Configuration.get(key) if Configuration.exists(key) else {}
try:
with remote(client.ip, [Migrator]) as rem:
rem.Migrator.migrate(master_ips, extra_ips)
except EOFError as eof:
UpdateController._logger.warning('{0}: EOFError during code migration, retrying {1}'.format(client.ip, eof))
with remote(client.ip, [Migrator]) as rem:
rem.Migrator.migrate(master_ips, extra_ips)
new_versions = Configuration.get(key) if Configuration.exists(key) else {}
if old_versions != new_versions:
UpdateController._logger.debug('{0}: Finished extensions code migration. Old versions: {1} --> New versions: {2}'.format(client.ip, old_versions, new_versions))
except Exception as ex:
failures.append('{0}: {1}'.format(client.ip, str(ex)))
if len(failures) > 0:
raise Exception('Failed to run the extensions migrate code on all nodes. Errors found:\n\n{0}'.format('\n\n'.join(failures)))
# Start memcached
if 'memcached' in services_stop_start:
services_stop_start.remove('memcached')
UpdateController._logger.debug('Starting memcached')
UpdateController.change_services_state(services=['memcached'],
ssh_clients=ssh_clients,
action='start')
# Migrate model
if 'framework' in components:
UpdateController._logger.debug('Verifying DAL code migration is required')
old_versions = PersistentFactory.get_client().get('ovs_model_version') if PersistentFactory.get_client().exists('ovs_model_version') else {}
from ovs.dal.helpers import Migration
with remote(ssh_clients[0].ip, [Migration]) as rem:
rem.Migration.migrate()
new_versions = PersistentFactory.get_client().get('ovs_model_version') if PersistentFactory.get_client().exists('ovs_model_version') else {}
if old_versions != new_versions:
UpdateController._logger.debug('Finished DAL code migration. Old versions: {0} --> New versions: {1}'.format(old_versions, new_versions))
# Post update actions
for client in ssh_clients:
UpdateController._logger.debug('{0}: Executing post-update actions'.format(client.ip))
for function in Toolbox.fetch_hooks('update', 'post_update_multi'):
try:
function(client=client, components=components)
except Exception as ex:
UpdateController._logger.exception('{0}: Post update hook {1} failed with error: {2}'.format(client.ip, function.__name__, ex))
for function in Toolbox.fetch_hooks('update', 'post_update_single'):
try:
function(components=components)
except Exception as ex:
UpdateController._logger.exception('Post update hook {0} failed with error: {1}'.format(function.__name__, ex))
# Start services
UpdateController.change_services_state(services=services_stop_start,
ssh_clients=ssh_clients,
action='start')
UpdateController._refresh_package_information()
UpdateController._logger.debug('+++ Finished updating +++')
except NoLockAvailableException:
UpdateController._logger.debug('Another update is currently in progress!')
except Exception as ex:
UpdateController._logger.exception('Error during update: {0}'.format(ex))
if len(ssh_clients) > 0:
UpdateController.change_services_state(services=services_stop_start,
ssh_clients=ssh_clients,
action='start')
UpdateController._refresh_package_information()
UpdateController._logger.error('Failed to update. Please check all the logs for more information')
finally:
filemutex.release()
for ssh_client in ssh_clients:
for file_name in [UpdateController._update_file, UpdateController._update_ongoing_file]:
try:
if ssh_client.file_exists(file_name):
ssh_client.file_delete(file_name)
except:
UpdateController._logger.warning('[0}: Failed to remove lock file {1}'.format(ssh_client.ip, file_name))
def update_volumedriver():
"""
Update the volumedriver
:return: None
"""
filemutex = file_mutex('system_update', wait=2)
upgrade_file = '/etc/ready_for_upgrade'
upgrade_ongoing_check_file = '/etc/upgrade_ongoing'
ssh_clients = []
try:
filemutex.acquire()
UpdateController._log_message('+++ Starting volumedriver update +++')
from ovs.dal.lists.storagerouterlist import StorageRouterList
UpdateController._log_message('Generating SSH client connections for each storage router')
storage_routers = StorageRouterList.get_storagerouters()
ssh_clients = [SSHClient(storage_router.ip, 'root') for storage_router in storage_routers]
this_client = [client for client in ssh_clients if client.is_local is True][0]
# Commence update !!!!!!!
# 0. Create locks
UpdateController._log_message('Creating lock files', client_ip=this_client.ip)
for client in ssh_clients:
client.run('touch {0}'.format(upgrade_file)) # Prevents manual install or upgrade individual packages
client.run('touch {0}'.format(upgrade_ongoing_check_file)) # Prevents clicking x times on 'Update' btn
# 1. Check requirements
packages_to_update = set()
all_services_to_restart = []
for client in ssh_clients:
for function in Toolbox.fetch_hooks('update', 'metadata'):
UpdateController._log_message('Executing function {0}'.format(function.__name__),
client_ip=client.ip)
output = function(client)
for key, value in output.iteritems():
if key != 'volumedriver':
continue
for package_info in value:
packages_to_update.update(package_info['packages'])
all_services_to_restart += package_info['services']
services_to_restart = []
for service in all_services_to_restart:
if service not in services_to_restart:
services_to_restart.append(service) # Filter out duplicates keeping the order of services (eg: watcher-framework before memcached)
UpdateController._log_message('Services which will be restarted --> {0}'.format(', '.join(services_to_restart)))
UpdateController._log_message('Packages which will be installed --> {0}'.format(', '.join(packages_to_update)))
# 1. Stop services
if UpdateController._change_services_state(services=services_to_restart,
ssh_clients=ssh_clients,
action='stop') is False:
UpdateController._log_message('Stopping all services on every node failed, cannot continue',
client_ip=this_client.ip, severity='warning')
UpdateController._remove_lock_files([upgrade_file, upgrade_ongoing_check_file], ssh_clients)
UpdateController._log_message('Attempting to start the services again', client_ip=this_client.ip)
UpdateController._change_services_state(services=services_to_restart,
ssh_clients=ssh_clients,
action='start')
UpdateController._log_message('Failed to stop all required services, update aborted',
client_ip=this_client.ip, severity='error')
return
# 2. Update packages
failed_clients = []
for client in ssh_clients:
PackageManager.update(client=client)
try:
for package_name in packages_to_update:
UpdateController._log_message('Installing {0}'.format(package_name), client.ip)
PackageManager.install(package_name=package_name,
client=client,
force=True)
UpdateController._log_message('Installed {0}'.format(package_name), client.ip)
client.file_delete(upgrade_file)
except subprocess.CalledProcessError as cpe:
UpdateController._log_message('Upgrade failed with error: {0}'.format(cpe.output), client.ip,
'error')
failed_clients.append(client)
break
if failed_clients:
UpdateController._remove_lock_files([upgrade_file, upgrade_ongoing_check_file], ssh_clients)
UpdateController._log_message('Error occurred. Attempting to start all services again',
client_ip=this_client.ip, severity='error')
UpdateController._change_services_state(services=services_to_restart,
ssh_clients=ssh_clients,
action='start')
UpdateController._log_message('Failed to upgrade following nodes:\n - {0}\nPlease check /var/log/ovs/lib.log on {1} for more information'.format('\n - '.join([client.ip for client in failed_clients]), this_client.ip),
this_client.ip,
'error')
return
# 3. Post upgrade actions
UpdateController._log_message('Executing post upgrade actions', client_ip=this_client.ip)
for client in ssh_clients:
for function in Toolbox.fetch_hooks('update', 'postupgrade'):
UpdateController._log_message('Executing action: {0}'.format(function.__name__), client_ip=client.ip)
try:
function(client)
except Exception as ex:
UpdateController._log_message('Post upgrade action failed with error: {0}'.format(ex),
client.ip, 'error')
# 4. Start services
UpdateController._log_message('Starting services', client_ip=this_client.ip)
UpdateController._change_services_state(services=services_to_restart,
ssh_clients=ssh_clients,
action='start')
UpdateController._remove_lock_files([upgrade_ongoing_check_file], ssh_clients)
UpdateController._log_message('+++ Finished updating +++')
except RuntimeError as rte:
UpdateController._log_message('Error during volumedriver update: {0}'.format(rte), severity='error')
UpdateController._remove_lock_files([upgrade_file, upgrade_ongoing_check_file], ssh_clients)
except NoLockAvailableException:
UpdateController._log_message('Another volumedriver update is currently in progress!')
except Exception as ex:
UpdateController._log_message('Error during volumedriver update: {0}'.format(ex), severity='error')
UpdateController._remove_lock_files([upgrade_file, upgrade_ongoing_check_file], ssh_clients)
finally:
filemutex.release()
def refresh_package_information():
"""
Retrieve and store the package information of all StorageRouters
:return: None
"""
GenericController._logger.info('Updating package information')
client_map = {}
prerequisites = []
package_info_cluster = {}
all_storagerouters = StorageRouterList.get_storagerouters()
all_storagerouters.sort(key=lambda sr: ExtensionsToolbox.advanced_sort(
element=sr.ip, separator='.'))
for storagerouter in all_storagerouters:
package_info_cluster[storagerouter.ip] = {}
try:
# We make use of these clients in Threads --> cached = False
client_map[storagerouter] = SSHClient(endpoint=storagerouter,
username='******',
cached=False)
except (NotAuthenticatedException, UnableToConnectException):
GenericController._logger.warning(
'StorageRouter {0} is inaccessible'.format(
storagerouter.ip))
prerequisites.append(['node_down', storagerouter.name])
package_info_cluster[storagerouter.ip]['errors'] = [
'StorageRouter {0} is inaccessible'.format(
storagerouter.name)
]
# Retrieve for each StorageRouter in the cluster the installed and candidate versions of related packages
# This also validates whether all required packages have been installed
GenericController._logger.debug(
'Retrieving package information for the cluster')
threads = []
for storagerouter, client in client_map.iteritems():
for fct in Toolbox.fetch_hooks(
component='update',
sub_component='get_package_update_info_cluster'):
thread = Thread(target=fct,
args=(client, package_info_cluster))
thread.start()
threads.append(thread)
for thread in threads:
thread.join()
# Retrieve the related downtime / service restart information
GenericController._logger.debug(
'Retrieving update information for the cluster')
update_info_cluster = {}
for storagerouter, client in client_map.iteritems():
update_info_cluster[storagerouter.ip] = {
'errors':
package_info_cluster[storagerouter.ip].get('errors', [])
}
for fct in Toolbox.fetch_hooks(
component='update',
sub_component='get_update_info_cluster'):
fct(client, update_info_cluster,
package_info_cluster[storagerouter.ip])
# Retrieve the update information for plugins (eg: ALBA, iSCSI)
GenericController._logger.debug(
'Retrieving package and update information for the plugins')
threads = []
update_info_plugin = {}
for fct in Toolbox.fetch_hooks('update', 'get_update_info_plugin'):
thread = Thread(target=fct, args=(update_info_plugin, ))
thread.start()
threads.append(thread)
for thread in threads:
thread.join()
# Add the prerequisites
if len(prerequisites) > 0:
for ip, component_info in update_info_cluster.iteritems():
if PackageFactory.COMP_FWK in component_info:
component_info[PackageFactory.COMP_FWK][
'prerequisites'].extend(prerequisites)
# Store information in model and collect errors for OVS cluster
errors = set()
for storagerouter in all_storagerouters:
GenericController._logger.debug(
'Storing update information for StorageRouter {0}'.format(
storagerouter.ip))
update_info = update_info_cluster.get(storagerouter.ip, {})
# Remove the errors from the update information
sr_errors = update_info.pop('errors', [])
if len(sr_errors) > 0:
errors.update([
'{0}: {1}'.format(storagerouter.ip, error)
for error in sr_errors
])
update_info = {
} # If any error occurred, we store no update information for this StorageRouter
# Remove the components without updates from the update information
update_info_copy = copy.deepcopy(update_info)
for component, info in update_info_copy.iteritems():
if len(info['packages']) == 0:
update_info.pop(component)
# Store the update information
storagerouter.package_information = update_info
storagerouter.save()
# Collect errors for plugins
for ip, plugin_errors in update_info_plugin.iteritems():
if len(plugin_errors) > 0:
errors.update(
['{0}: {1}'.format(ip, error) for error in plugin_errors])
if len(errors) > 0:
raise Exception('\n - {0}'.format('\n - '.join(errors)))
GenericController._logger.info('Finished updating package information')
def remove_node(node_ip, silent=None):
"""
Remove the node with specified IP from the cluster
:param node_ip: IP of the node to remove
:type node_ip: str
:param silent: If silent == '--force-yes' no question will be asked to confirm the removal
:type silent: str
:return: None
"""
from ovs.lib.storagedriver import StorageDriverController
from ovs.lib.storagerouter import StorageRouterController
from ovs.dal.lists.storagerouterlist import StorageRouterList
Toolbox.log(logger=NodeRemovalController._logger, messages="Remove node", boxed=True)
Toolbox.log(
logger=NodeRemovalController._logger,
messages="WARNING: Some of these steps may take a very long time, please check the logs for more information\n\n",
)
###############
# VALIDATIONS #
###############
try:
node_ip = node_ip.strip()
if not isinstance(node_ip, str):
raise ValueError("Node IP must be a string")
if not re.match(SSHClient.IP_REGEX, node_ip):
raise ValueError("Invalid IP {0} specified".format(node_ip))
storage_router_all = StorageRouterList.get_storagerouters()
storage_router_masters = StorageRouterList.get_masters()
storage_router_all_ips = set([storage_router.ip for storage_router in storage_router_all])
storage_router_master_ips = set([storage_router.ip for storage_router in storage_router_masters])
storage_router_to_remove = StorageRouterList.get_by_ip(node_ip)
if node_ip not in storage_router_all_ips:
raise ValueError(
"Unknown IP specified\nKnown in model:\n - {0}\nSpecified for removal:\n - {1}".format(
"\n - ".join(storage_router_all_ips), node_ip
)
)
if len(storage_router_all_ips) == 1:
raise RuntimeError("Removing the only node is not possible")
if node_ip in storage_router_master_ips and len(storage_router_master_ips) == 1:
raise RuntimeError("Removing the only master node is not possible")
if System.get_my_storagerouter() == storage_router_to_remove:
raise RuntimeError(
"The node to be removed cannot be identical to the node on which the removal is initiated"
)
Toolbox.log(
logger=NodeRemovalController._logger, messages="Creating SSH connections to remaining master nodes"
)
master_ip = None
ip_client_map = {}
storage_routers_offline = []
storage_router_to_remove_online = True
for storage_router in storage_router_all:
try:
client = SSHClient(storage_router, username="******")
if client.run(["pwd"]):
Toolbox.log(
logger=NodeRemovalController._logger,
messages=" Node with IP {0:<15} successfully connected to".format(storage_router.ip),
)
ip_client_map[storage_router.ip] = client
if storage_router != storage_router_to_remove and storage_router.node_type == "MASTER":
master_ip = storage_router.ip
except UnableToConnectException:
Toolbox.log(
logger=NodeRemovalController._logger,
messages=" Node with IP {0:<15} is unreachable".format(storage_router.ip),
)
storage_routers_offline.append(storage_router)
if storage_router == storage_router_to_remove:
storage_router_to_remove_online = False
if len(ip_client_map) == 0 or master_ip is None:
raise RuntimeError("Could not connect to any master node in the cluster")
storage_router_to_remove.invalidate_dynamics("vdisks_guids")
if (
len(storage_router_to_remove.vdisks_guids) > 0
): # vDisks are supposed to be moved away manually before removing a node
raise RuntimeError("Still vDisks attached to Storage Router {0}".format(storage_router_to_remove.name))
internal_memcached = Toolbox.is_service_internally_managed(service="memcached")
internal_rabbit_mq = Toolbox.is_service_internally_managed(service="rabbitmq")
memcached_endpoints = Configuration.get(key="/ovs/framework/memcache|endpoints")
rabbit_mq_endpoints = Configuration.get(key="/ovs/framework/messagequeue|endpoints")
copy_memcached_endpoints = list(memcached_endpoints)
copy_rabbit_mq_endpoints = list(rabbit_mq_endpoints)
for endpoint in memcached_endpoints:
if endpoint.startswith(storage_router_to_remove.ip):
copy_memcached_endpoints.remove(endpoint)
for endpoint in rabbit_mq_endpoints:
if endpoint.startswith(storage_router_to_remove.ip):
copy_rabbit_mq_endpoints.remove(endpoint)
if len(copy_memcached_endpoints) == 0 and internal_memcached is True:
raise RuntimeError(
"Removal of provided nodes will result in a complete removal of the memcached service"
)
if len(copy_rabbit_mq_endpoints) == 0 and internal_rabbit_mq is True:
raise RuntimeError(
"Removal of provided nodes will result in a complete removal of the messagequeue service"
)
except Exception as exception:
Toolbox.log(
logger=NodeRemovalController._logger, messages=[str(exception)], boxed=True, loglevel="exception"
)
sys.exit(1)
#################
# CONFIRMATIONS #
#################
interactive = silent != "--force-yes"
remove_asd_manager = not interactive # Remove ASD manager if non-interactive else ask
if interactive is True:
proceed = Interactive.ask_yesno(
message="Are you sure you want to remove node {0}?".format(storage_router_to_remove.name),
default_value=False,
)
if proceed is False:
Toolbox.log(logger=NodeRemovalController._logger, messages="Abort removal", title=True)
sys.exit(1)
if storage_router_to_remove_online is True:
client = SSHClient(endpoint=storage_router_to_remove, username="******")
if ServiceManager.has_service(name="asd-manager", client=client):
remove_asd_manager = Interactive.ask_yesno(
message="Do you also want to remove the ASD manager and related ASDs?", default_value=False
)
if remove_asd_manager is True or storage_router_to_remove_online is False:
for function in Toolbox.fetch_hooks("setup", "validate_asd_removal"):
validation_output = function(storage_router_to_remove.ip)
if validation_output["confirm"] is True:
if Interactive.ask_yesno(message=validation_output["question"], default_value=False) is False:
remove_asd_manager = False
break
###########
# REMOVAL #
###########
try:
Toolbox.log(
logger=NodeRemovalController._logger,
messages="Starting removal of node {0} - {1}".format(
storage_router_to_remove.name, storage_router_to_remove.ip
),
)
if storage_router_to_remove_online is False:
Toolbox.log(
logger=NodeRemovalController._logger,
messages=" Marking all Storage Drivers served by Storage Router {0} as offline".format(
storage_router_to_remove.ip
),
)
StorageDriverController.mark_offline(storagerouter_guid=storage_router_to_remove.guid)
# Remove vPools
Toolbox.log(
logger=NodeRemovalController._logger,
messages=" Removing vPools from node".format(storage_router_to_remove.ip),
)
storage_routers_offline_guids = [
sr.guid for sr in storage_routers_offline if sr.guid != storage_router_to_remove.guid
]
for storage_driver in storage_router_to_remove.storagedrivers:
Toolbox.log(
logger=NodeRemovalController._logger,
messages=" Removing vPool {0} from node".format(storage_driver.vpool.name),
)
StorageRouterController.remove_storagedriver(
storagedriver_guid=storage_driver.guid, offline_storage_router_guids=storage_routers_offline_guids
)
# Demote if MASTER
if storage_router_to_remove.node_type == "MASTER":
NodeTypeController.demote_node(
cluster_ip=storage_router_to_remove.ip,
master_ip=master_ip,
ip_client_map=ip_client_map,
unique_id=storage_router_to_remove.machine_id,
unconfigure_memcached=internal_memcached,
unconfigure_rabbitmq=internal_rabbit_mq,
offline_nodes=storage_routers_offline,
)
# Stop / remove services
Toolbox.log(logger=NodeRemovalController._logger, messages="Stopping and removing services")
config_store = Configuration.get_store()
if storage_router_to_remove_online is True:
client = SSHClient(endpoint=storage_router_to_remove, username="******")
NodeRemovalController.remove_services(
client=client,
node_type=storage_router_to_remove.node_type.lower(),
logger=NodeRemovalController._logger,
)
service = "watcher-config"
if ServiceManager.has_service(service, client=client):
Toolbox.log(logger=NodeRemovalController._logger, messages="Removing service {0}".format(service))
ServiceManager.stop_service(service, client=client)
ServiceManager.remove_service(service, client=client)
if config_store == "etcd":
from ovs.extensions.db.etcd.installer import EtcdInstaller
if Configuration.get(key="/ovs/framework/external_config") is None:
Toolbox.log(logger=NodeRemovalController._logger, messages=" Removing Etcd cluster")
try:
EtcdInstaller.stop("config", client)
EtcdInstaller.remove("config", client)
except Exception as ex:
Toolbox.log(
logger=NodeRemovalController._logger,
messages=["\nFailed to unconfigure Etcd", ex],
loglevel="exception",
)
Toolbox.log(logger=NodeRemovalController._logger, messages="Removing Etcd proxy")
EtcdInstaller.remove_proxy("config", client.ip)
Toolbox.run_hooks(
component="noderemoval",
sub_component="remove",
logger=NodeRemovalController._logger,
cluster_ip=storage_router_to_remove.ip,
complete_removal=remove_asd_manager,
)
# Clean up model
Toolbox.log(logger=NodeRemovalController._logger, messages="Removing node from model")
for service in storage_router_to_remove.services:
service.delete()
for disk in storage_router_to_remove.disks:
for partition in disk.partitions:
partition.delete()
disk.delete()
for j_domain in storage_router_to_remove.domains:
j_domain.delete()
Configuration.delete("/ovs/framework/hosts/{0}".format(storage_router_to_remove.machine_id))
NodeTypeController.restart_framework_and_memcache_services(
clients=ip_client_map,
offline_node_ips=[node.ip for node in storage_routers_offline],
logger=NodeRemovalController._logger,
)
if storage_router_to_remove_online is True:
client = SSHClient(endpoint=storage_router_to_remove, username="******")
if config_store == "arakoon":
client.file_delete(filenames=[ArakoonConfiguration.CACC_LOCATION])
client.file_delete(filenames=[Configuration.BOOTSTRAP_CONFIG_LOCATION])
storage_router_to_remove.delete()
Toolbox.log(logger=NodeRemovalController._logger, messages="Successfully removed node\n")
except Exception as exception:
Toolbox.log(logger=NodeRemovalController._logger, messages="\n")
Toolbox.log(
logger=NodeRemovalController._logger,
messages=["An unexpected error occurred:", str(exception)],
boxed=True,
loglevel="exception",
)
sys.exit(1)
except KeyboardInterrupt:
Toolbox.log(logger=NodeRemovalController._logger, messages="\n")
Toolbox.log(
logger=NodeRemovalController._logger,
messages="This setup was aborted. Open vStorage may be in an inconsistent state, make sure to validate the installation.",
boxed=True,
loglevel="error",
)
sys.exit(1)
if remove_asd_manager is True:
Toolbox.log(logger=NodeRemovalController._logger, messages="\nRemoving ASD Manager")
with remote(storage_router_to_remove.ip, [os]) as rem:
rem.os.system("asd-manager remove --force-yes")
Toolbox.log(logger=NodeRemovalController._logger, messages="Remove nodes finished", title=True)
def _can_remove(self):
"""
Can be removed
"""
return len(Toolbox.fetch_hooks('license', '{0}.remove'.format(self.component))) == 1
