def get(pmachine):
"""
Returns the appropriate hypervisor client class for a given PMachine
"""
hvtype = pmachine.hvtype
ip = pmachine.ip
username = pmachine.username
password = pmachine.password
key = '{0}_{1}'.format(ip, username)
if key not in Factory.hypervisors:
mutex = FileMutex('hypervisor_{0}'.format(key))
try:
mutex.acquire(30)
if key not in Factory.hypervisors:
if hvtype == 'VMWARE':
from hypervisors.vmware import VMware
hypervisor = VMware(ip, username, password)
elif hvtype == 'KVM':
from hypervisors.kvm import KVM
hypervisor = KVM(ip, username, password)
else:
raise NotImplementedError('Hypervisor {0} is not yet supported'.format(hvtype))
Factory.hypervisors[key] = hypervisor
finally:
mutex.release()
return Factory.hypervisors[key]
def get_mgmtcenter(pmachine=None, mgmt_center=None):
"""
@param pmachine: pmachine hybrid from DAL
@param mgmt_center: mgmtcenter hybrid from DAL
Returns the appropriate sdk client for the management center of the node
"""
if not ((pmachine is None) ^ (mgmt_center is None)):
raise ValueError('Either a pMachine or a Management center should be passed')
if pmachine is not None:
mgmt_center = pmachine.mgmtcenter
if mgmt_center is None:
return None
mgmtcenter_type = mgmt_center.type
ip = mgmt_center.ip
username = mgmt_center.username
password = mgmt_center.password
key = '{0}_{1}'.format(ip, username)
if key not in Factory.mgmtcenters:
mutex = FileMutex('mgmtcenter_{0}'.format(key))
try:
mutex.acquire(30)
if key not in Factory.mgmtcenters:
if mgmtcenter_type == 'VCENTER':
from mgmtcenters.vcenter import VCenter
mgmtcenter = VCenter(ip, username, password)
elif mgmtcenter_type == 'OPENSTACK':
from mgmtcenters.openstack import OpenStack
mgmtcenter = OpenStack(ip, username, password)
else:
raise NotImplementedError('Management center for {0} is not yet supported'.format(mgmtcenter_type))
Factory.mgmtcenters[key] = mgmtcenter
finally:
mutex.release()
return Factory.mgmtcenters[key]
def new_function(*args, **kw):
"""
Executes the decorated function in a locked context
"""
filemutex = FileMutex('messaging')
try:
filemutex.acquire(wait=5)
mutex = VolatileMutex('messaging')
try:
mutex.acquire(wait=5)
return f(*args, **kw)
finally:
mutex.release()
finally:
filemutex.release()
def get(pmachine):
"""
Returns the appropriate hypervisor client class for a given PMachine
"""
hvtype = pmachine.hvtype
ip = pmachine.ip
username = pmachine.username
password = pmachine.password
key = '{0}_{1}'.format(ip, username)
if key not in Factory.hypervisors:
mutex = FileMutex('hypervisor_{0}'.format(key))
try:
mutex.acquire(30)
if key not in Factory.hypervisors:
if hvtype == 'VMWARE':
from hypervisors.vmware import VMware
hypervisor = VMware(ip, username, password)
elif hvtype == 'KVM':
from hypervisors.kvm import KVM
hypervisor = KVM(ip, username, password)
else:
raise NotImplementedError(
'Hypervisor {0} is not yet supported'.format(
hvtype))
Factory.hypervisors[key] = hypervisor
finally:
mutex.release()
return Factory.hypervisors[key]
def new_function(*args, **kw):
"""
Executes the decorated function in a locked context
"""
filemutex = FileMutex('messaging')
try:
filemutex.acquire(wait=5)
mutex = VolatileMutex('messaging')
try:
mutex.acquire(wait=5)
return f(*args, **kw)
finally:
mutex.release()
finally:
filemutex.release()
def get_mgmtcenter(pmachine=None, mgmt_center=None):
"""
@param pmachine: pmachine hybrid from DAL
@param mgmt_center: mgmtcenter hybrid from DAL
Returns the appropriate sdk client for the management center of the node
Implemented only for VMWare / vCenter Server
TODO: KVM
"""
if not ((pmachine is None) ^ (mgmt_center is None)):
raise ValueError(
'Either a pMachine or a Management center should be passed')
if pmachine is not None:
mgmt_center = pmachine.mgmtcenter
if mgmt_center is None:
return None
mgmtcenter_type = mgmt_center.type
ip = mgmt_center.ip
username = mgmt_center.username
password = mgmt_center.password
key = '{0}_{1}'.format(ip, username)
if key not in Factory.mgmtcenters:
mutex = FileMutex('mgmtcenter_{0}'.format(key))
try:
mutex.acquire(30)
if key not in Factory.mgmtcenters:
if mgmtcenter_type == 'VCENTER':
from mgmtcenters.vcenter import VCenter
mgmtcenter = VCenter(ip, username, password)
elif mgmtcenter_type == 'OPENSTACK':
from mgmtcenters.openstack import OpenStack
mgmtcenter = OpenStack(ip, username, password)
else:
raise NotImplementedError(
'Management center for {0} is not yet supported'.
format(mgmtcenter_type))
Factory.mgmtcenters[key] = mgmtcenter
finally:
mutex.release()
return Factory.mgmtcenters[key]
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:
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:
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:
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:
file_mutex.release()
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()
