¡@
OpenStack Study: nsx_utils.py**** CubicPower OpenStack Study **** def fetch_nsx_switches(session, cluster, neutron_net_id): """Retrieve logical switches for a neutron network. This function is optimized for fetching all the lswitches always with a single NSX query. If there is more than 1 logical switch (chained switches use case) NSX lswitches are queried by 'quantum_net_id' tag. Otherwise the NSX lswitch is directly retrieved by id (more efficient). """ nsx_switch_ids = get_nsx_switch_ids(session, cluster, neutron_net_id) if len(nsx_switch_ids) > 1: lswitches = switchlib.get_lswitches(cluster, neutron_net_id) else: lswitches = [switchlib.get_lswitch_by_id( cluster, nsx_switch_ids[0])] return lswitches
**** CubicPower OpenStack Study **** def get_nsx_switch_ids(session, cluster, neutron_network_id): """Return the NSX switch id for a given neutron network. First lookup for mappings in Neutron database. If no mapping is found, query the NSX backend and add the mappings. """ nsx_switch_ids = nsx_db.get_nsx_switch_ids( session, neutron_network_id) if not nsx_switch_ids: # Find logical switches from backend. # This is a rather expensive query, but it won't be executed # more than once for each network in Neutron's lifetime nsx_switches = switchlib.get_lswitches(cluster, neutron_network_id) if not nsx_switches: LOG.warn(_("Unable to find NSX switches for Neutron network %s"), neutron_network_id) return nsx_switch_ids = [] with session.begin(subtransactions=True): for nsx_switch in nsx_switches: nsx_switch_id = nsx_switch['uuid'] nsx_switch_ids.append(nsx_switch_id) # Create DB mapping nsx_db.add_neutron_nsx_network_mapping( session, neutron_network_id, nsx_switch_id) return nsx_switch_ids
**** CubicPower OpenStack Study **** def get_nsx_switch_and_port_id(session, cluster, neutron_port_id): """Return the NSX switch and port uuids for a given neutron port. First, look up the Neutron database. If not found, execute a query on NSX platform as the mapping might be missing because the port was created before upgrading to grizzly. This routine also retrieves the identifier of the logical switch in the backend where the port is plugged. Prior to Icehouse this information was not available in the Neutron Database. For dealing with pre-existing records, this routine will query the backend for retrieving the correct switch identifier. As of Icehouse release it is not indeed anymore possible to assume the backend logical switch identifier is equal to the neutron network identifier. """ nsx_switch_id, nsx_port_id = nsx_db.get_nsx_switch_and_port_id( session, neutron_port_id) if not nsx_switch_id: # Find logical switch for port from backend # This is a rather expensive query, but it won't be executed # more than once for each port in Neutron's lifetime nsx_ports = switchlib.query_lswitch_lports( cluster, '*', relations='LogicalSwitchConfig', filters={'tag': neutron_port_id, 'tag_scope': 'q_port_id'}) # Only one result expected # NOTE(salv-orlando): Not handling the case where more than one # port is found with the same neutron port tag if not nsx_ports: LOG.warn(_("Unable to find NSX port for Neutron port %s"), neutron_port_id) # This method is supposed to return a tuple return None, None nsx_port = nsx_ports[0] nsx_switch_id = (nsx_port['_relations'] ['LogicalSwitchConfig']['uuid']) if nsx_port_id: # Mapping already exists. Delete before recreating nsx_db.delete_neutron_nsx_port_mapping( session, neutron_port_id) else: nsx_port_id = nsx_port['uuid'] # (re)Create DB mapping nsx_db.add_neutron_nsx_port_mapping( session, neutron_port_id, nsx_switch_id, nsx_port_id) return nsx_switch_id, nsx_port_id
**** CubicPower OpenStack Study **** def get_nsx_security_group_id(session, cluster, neutron_id): """Return the NSX sec profile uuid for a given neutron sec group. First, look up the Neutron database. If not found, execute a query on NSX platform as the mapping might be missing. NOTE: Security groups are called 'security profiles' on the NSX backend. """ nsx_id = nsx_db.get_nsx_security_group_id(session, neutron_id) if not nsx_id: # Find security profile on backend. # This is a rather expensive query, but it won't be executed # more than once for each security group in Neutron's lifetime nsx_sec_profiles = secgrouplib.query_security_profiles( cluster, '*', filters={'tag': neutron_id, 'tag_scope': 'q_sec_group_id'}) # Only one result expected # NOTE(salv-orlando): Not handling the case where more than one # security profile is found with the same neutron port tag if not nsx_sec_profiles: LOG.warn(_("Unable to find NSX security profile for Neutron " "security group %s"), neutron_id) return elif len(nsx_sec_profiles) > 1: LOG.warn(_("Multiple NSX security profiles found for Neutron " "security group %s"), neutron_id) nsx_sec_profile = nsx_sec_profiles[0] nsx_id = nsx_sec_profile['uuid'] with session.begin(subtransactions=True): # Create DB mapping nsx_db.add_neutron_nsx_security_group_mapping( session, neutron_id, nsx_id) return nsx_id
**** CubicPower OpenStack Study **** def get_nsx_router_id(session, cluster, neutron_router_id): """Return the NSX router uuid for a given neutron router. First, look up the Neutron database. If not found, execute a query on NSX platform as the mapping might be missing. """ nsx_router_id = nsx_db.get_nsx_router_id( session, neutron_router_id) if not nsx_router_id: # Find logical router from backend. # This is a rather expensive query, but it won't be executed # more than once for each router in Neutron's lifetime nsx_routers = routerlib.query_lrouters( cluster, '*', filters={'tag': neutron_router_id, 'tag_scope': 'q_router_id'}) # Only one result expected # NOTE(salv-orlando): Not handling the case where more than one # port is found with the same neutron port tag if not nsx_routers: LOG.warn(_("Unable to find NSX router for Neutron router %s"), neutron_router_id) return nsx_router = nsx_routers[0] nsx_router_id = nsx_router['uuid'] with session.begin(subtransactions=True): # Create DB mapping nsx_db.add_neutron_nsx_router_mapping( session, neutron_router_id, nsx_router_id) return nsx_router_id
**** CubicPower OpenStack Study **** def create_nsx_cluster(cluster_opts, concurrent_connections, gen_timeout): cluster = nsx_cluster.NSXCluster(**cluster_opts) def _ctrl_split(x, y): return (x, int(y), True) api_providers = [_ctrl_split(*ctrl.split(':')) for ctrl in cluster.nsx_controllers] cluster.api_client = client.NsxApiClient( api_providers, cluster.nsx_user, cluster.nsx_password, request_timeout=cluster.req_timeout, http_timeout=cluster.http_timeout, retries=cluster.retries, redirects=cluster.redirects, concurrent_connections=concurrent_connections, gen_timeout=gen_timeout) return cluster
**** CubicPower OpenStack Study **** def get_nsx_device_status(cluster, nsx_uuid): try: status_up = l2gwlib.get_gateway_device_status( cluster, nsx_uuid) if status_up: return networkgw_db.STATUS_ACTIVE else: return networkgw_db.STATUS_DOWN except api_exc.NsxApiException: return networkgw_db.STATUS_UNKNOWN except n_exc.NotFound: return networkgw_db.ERROR
**** CubicPower OpenStack Study **** def get_nsx_device_statuses(cluster, tenant_id): try: status_dict = l2gwlib.get_gateway_devices_status( cluster, tenant_id) return dict((nsx_device_id, networkgw_db.STATUS_ACTIVE if connected else networkgw_db.STATUS_DOWN) for (nsx_device_id, connected) in status_dict.iteritems()) except api_exc.NsxApiException: # Do not make a NSX API exception fatal if tenant_id: LOG.warn(_("Unable to retrieve operational status for gateway " "devices belonging to tenant: %s"), tenant_id) else: LOG.warn(_("Unable to retrieve operational status for " "gateway devices"))
|