async def check_RO_version(self):
tries = 14
last_error = None
while True:
try:
if self.config["ro_config"].get("ng"):
ro_server = NgRoClient(self.loop,
**self.config["ro_config"])
else:
ro_server = ROClient(self.loop, **self.config["ro_config"])
ro_version = await ro_server.get_version()
if versiontuple(ro_version) < versiontuple(min_RO_version):
raise LcmException(
"Not compatible osm/RO version '{}'. Needed '{}' or higher"
.format(ro_version, min_RO_version))
self.logger.info(
"Connected to RO version {}".format(ro_version))
return
except (ROClientException, NgRoException) as e:
tries -= 1
error_text = "Error while connecting to RO on {}: {}".format(
self.config["ro_config"]["uri"], e)
if tries <= 0:
self.logger.critical(error_text)
raise LcmException(error_text)
if last_error != error_text:
last_error = error_text
self.logger.error(
error_text +
". Waiting until {} seconds".format(5 * tries))
await asyncio.sleep(5)
async def send_data(self, new_config):
restore_backup = False
del new_config["_id"]
del new_config["_admin"]
new_scrape_configs = []
# generate a list with the values of scrape_configs
for scrape_config in new_config["scrape_configs"].values():
scrape_config = scrape_config.copy()
# remove nsr_id metadata from scrape_configs
scrape_config.pop("nsr_id", None)
new_scrape_configs.append(scrape_config)
new_config["scrape_configs"] = new_scrape_configs
try:
if os.path.exists(self.cfg_file):
os.rename(self.cfg_file, self.cfg_file_backup)
restore_backup = True
with open(self.cfg_file, "w+") as f:
yaml.safe_dump(new_config,
f,
indent=4,
default_flow_style=False)
# self.logger.debug("new configuration: {}".format(yaml.safe_dump(new_config, indent=4,
# default_flow_style=False)))
async with aiohttp.ClientSession() as session:
async with session.post(self.server + "-/reload") as resp:
if resp.status > 204:
raise LcmException(await resp.text())
await asyncio.sleep(5, loop=self.loop)
# If prometheus does not admit this configuration, remains with the old one
# Then, to check if the configuration has been accepted, get the configuration from prometheus
# and compares with the inserted one
async with session.get(self.server +
"api/v1/status/config") as resp:
if resp.status > 204:
raise LcmException(await resp.text())
current_config = await resp.json()
if not self._check_configuration_equal(
current_config, new_config):
return False
else:
restore_backup = False
return True
except Exception as e:
self.logger.error("Error updating configuration url={}: {}".format(
self.server, e))
return False
finally:
if restore_backup:
try:
os.rename(self.cfg_file_backup, self.cfg_file)
except Exception as e:
self.logger.critical(
"Exception while rolling back: {}".format(e))
def parse_job(job_data: str, variables: dict) -> dict:
try:
template = Template(job_data)
job_parsed = template.render(variables or {})
return yaml.safe_load(job_parsed)
except (TemplateError, TemplateNotFound, TemplateSyntaxError) as e:
# TODO yaml exceptions
raise LcmException(
"Error parsing Jinja2 to prometheus job. job_data={}, variables={}. Error={}"
.format(job_data, variables, e))
def vim_account_2_RO(vim_account):
"""
Translate a RO vim_account from OSM vim_account params
:param ns_params: OSM instantiate params
:return: The RO ns descriptor
"""
if vim_account in vim_2_RO:
return vim_2_RO[vim_account]
db_vim = self.db.get_one("vim_accounts", {"_id": vim_account})
if db_vim["_admin"]["operationalState"] != "ENABLED":
raise LcmException(
"VIM={} is not available. operationalState={}".format(
vim_account, db_vim["_admin"]["operationalState"]))
RO_vim_id = db_vim["_admin"]["deployed"]["RO"]
vim_2_RO[vim_account] = RO_vim_id
return RO_vim_id
async def kafka_ping(self):
self.logger.debug("Task kafka_ping Enter")
consecutive_errors = 0
first_start = True
kafka_has_received = False
self.pings_not_received = 1
while True:
try:
await self.msg_admin.aiowrite(
"admin", "ping", {
"from": "lcm",
"to": "lcm",
"worker_id": self.worker_id,
"version": lcm_version
}, self.loop)
# time between pings are low when it is not received and at starting
wait_time = self.ping_interval_boot if not kafka_has_received else self.ping_interval_pace
if not self.pings_not_received:
kafka_has_received = True
self.pings_not_received += 1
await asyncio.sleep(wait_time, loop=self.loop)
if self.pings_not_received > 10:
raise LcmException(
"It is not receiving pings from Kafka bus")
consecutive_errors = 0
first_start = False
except LcmException:
raise
except Exception as e:
# if not first_start is the first time after starting. So leave more time and wait
# to allow kafka starts
if consecutive_errors == 8 if not first_start else 30:
self.logger.error(
"Task kafka_read task exit error too many errors. Exception: {}"
.format(e))
raise
consecutive_errors += 1
self.logger.error(
"Task kafka_read retrying after Exception {}".format(e))
wait_time = 2 if not first_start else 5
await asyncio.sleep(wait_time, loop=self.loop)
async def start(self):
for retry in range(4):
try:
# self.logger("Starting prometheus ")
# read from database
prometheus_data = self.db.get_one("admin",
{"_id": "prometheus"},
fail_on_empty=False)
if not prometheus_data:
self.logger.info("Init db.admin.prometheus content")
self.db.create("admin", initial_prometheus_data)
# send database config file to prometheus. Ignore loading errors, as prometheus may be starting
# but at least an initial configuration file is set
await self.update()
return
except DbException as e:
if retry == 3:
raise LcmException(
"Max retries trying to init prometheus configuration: {}"
.format(e))
await asyncio.sleep(5, loop=self.loop)
def nsi_update_nsir(self, nsi_update_nsir, db_nsir, nsir_desc_RO):
"""
Updates database nsir with the RO info for the created vld
:param nsi_update_nsir: dictionary to be filled with the updated info
:param db_nsir: content of db_nsir. This is also modified
:param nsir_desc_RO: nsir descriptor from RO
:return: Nothing, LcmException is raised on errors
"""
for vld_index, vld in enumerate(get_iterable(db_nsir, "vld")):
for net_RO in get_iterable(nsir_desc_RO, "nets"):
if vld["id"] != net_RO.get("ns_net_osm_id"):
continue
vld["vim-id"] = net_RO.get("vim_net_id")
vld["name"] = net_RO.get("vim_name")
vld["status"] = net_RO.get("status")
vld["status-detailed"] = net_RO.get("error_msg")
nsi_update_nsir["vld.{}".format(vld_index)] = vld
break
else:
raise LcmException(
"ns_update_nsir: Not found vld={} at RO info".format(
vld["id"]))
def __init__(self, config_file, loop=None):
"""
Init, Connect to database, filesystem storage, and messaging
:param config: two level dictionary with configuration. Top level should contain 'database', 'storage',
:return: None
"""
self.db = None
self.msg = None
self.msg_admin = None
self.fs = None
self.pings_not_received = 1
self.consecutive_errors = 0
self.first_start = False
# logging
self.logger = logging.getLogger('lcm')
# get id
self.worker_id = self.get_process_id()
# load configuration
config = self.read_config_file(config_file)
self.config = config
self.config["ro_config"] = {
"ng": config["RO"].get("ng", False),
"uri": config["RO"].get("uri"),
"tenant": config.get("tenant", "osm"),
"logger_name": "lcm.roclient",
"loglevel": config["RO"].get("loglevel", "ERROR"),
}
if not self.config["ro_config"]["uri"]:
if not self.config["ro_config"]["ng"]:
self.config["ro_config"][
"uri"] = "http://{}:{}/openmano".format(
config["RO"]["host"], config["RO"]["port"])
else:
self.config["ro_config"]["uri"] = "http://{}:{}/ro".format(
config["RO"]["host"], config["RO"]["port"])
self.loop = loop or asyncio.get_event_loop()
# logging
log_format_simple = "%(asctime)s %(levelname)s %(name)s %(filename)s:%(lineno)s %(message)s"
log_formatter_simple = logging.Formatter(log_format_simple,
datefmt='%Y-%m-%dT%H:%M:%S')
config["database"]["logger_name"] = "lcm.db"
config["storage"]["logger_name"] = "lcm.fs"
config["message"]["logger_name"] = "lcm.msg"
if config["global"].get("logfile"):
file_handler = logging.handlers.RotatingFileHandler(
config["global"]["logfile"],
maxBytes=100e6,
backupCount=9,
delay=0)
file_handler.setFormatter(log_formatter_simple)
self.logger.addHandler(file_handler)
if not config["global"].get("nologging"):
str_handler = logging.StreamHandler()
str_handler.setFormatter(log_formatter_simple)
self.logger.addHandler(str_handler)
if config["global"].get("loglevel"):
self.logger.setLevel(config["global"]["loglevel"])
# logging other modules
for k1, logname in self.cfg_logger_name.items():
config[k1]["logger_name"] = logname
logger_module = logging.getLogger(logname)
if config[k1].get("logfile"):
file_handler = logging.handlers.RotatingFileHandler(
config[k1]["logfile"],
maxBytes=100e6,
backupCount=9,
delay=0)
file_handler.setFormatter(log_formatter_simple)
logger_module.addHandler(file_handler)
if config[k1].get("loglevel"):
logger_module.setLevel(config[k1]["loglevel"])
self.logger.critical("starting osm/lcm version {} {}".format(
lcm_version, lcm_version_date))
# check version of N2VC
# TODO enhance with int conversion or from distutils.version import LooseVersion
# or with list(map(int, version.split(".")))
if versiontuple(n2vc_version) < versiontuple(min_n2vc_version):
raise LcmException(
"Not compatible osm/N2VC version '{}'. Needed '{}' or higher".
format(n2vc_version, min_n2vc_version))
# check version of common
if versiontuple(common_version) < versiontuple(min_common_version):
raise LcmException(
"Not compatible osm/common version '{}'. Needed '{}' or higher"
.format(common_version, min_common_version))
try:
# TODO check database version
if config["database"]["driver"] == "mongo":
self.db = dbmongo.DbMongo()
self.db.db_connect(config["database"])
elif config["database"]["driver"] == "memory":
self.db = dbmemory.DbMemory()
self.db.db_connect(config["database"])
else:
raise LcmException(
"Invalid configuration param '{}' at '[database]':'driver'"
.format(config["database"]["driver"]))
if config["storage"]["driver"] == "local":
self.fs = fslocal.FsLocal()
self.fs.fs_connect(config["storage"])
elif config["storage"]["driver"] == "mongo":
self.fs = fsmongo.FsMongo()
self.fs.fs_connect(config["storage"])
else:
raise LcmException(
"Invalid configuration param '{}' at '[storage]':'driver'".
format(config["storage"]["driver"]))
# copy message configuration in order to remove 'group_id' for msg_admin
config_message = config["message"].copy()
config_message["loop"] = self.loop
if config_message["driver"] == "local":
self.msg = msglocal.MsgLocal()
self.msg.connect(config_message)
self.msg_admin = msglocal.MsgLocal()
config_message.pop("group_id", None)
self.msg_admin.connect(config_message)
elif config_message["driver"] == "kafka":
self.msg = msgkafka.MsgKafka()
self.msg.connect(config_message)
self.msg_admin = msgkafka.MsgKafka()
config_message.pop("group_id", None)
self.msg_admin.connect(config_message)
else:
raise LcmException(
"Invalid configuration param '{}' at '[message]':'driver'".
format(config["message"]["driver"]))
except (DbException, FsException, MsgException) as e:
self.logger.critical(str(e), exc_info=True)
raise LcmException(str(e))
# contains created tasks/futures to be able to cancel
self.lcm_tasks = TaskRegistry(self.worker_id, self.db, self.logger)
if self.config.get("tsdb") and self.config["tsdb"].get("driver"):
if self.config["tsdb"]["driver"] == "prometheus":
self.prometheus = prometheus.Prometheus(
self.config["tsdb"], self.worker_id, self.db, self.loop)
else:
raise LcmException(
"Invalid configuration param '{}' at '[tsdb]':'driver'".
format(config["tsdb"]["driver"]))
else:
self.prometheus = None
self.ns = ns.NsLcm(self.db, self.msg, self.fs, self.lcm_tasks,
self.config, self.loop, self.prometheus)
self.netslice = netslice.NetsliceLcm(self.db, self.msg, self.fs,
self.lcm_tasks, self.config,
self.loop, self.ns)
self.vim = vim_sdn.VimLcm(self.db, self.msg, self.fs, self.lcm_tasks,
self.config, self.loop)
self.wim = vim_sdn.WimLcm(self.db, self.msg, self.fs, self.lcm_tasks,
self.config, self.loop)
self.sdn = vim_sdn.SdnLcm(self.db, self.msg, self.fs, self.lcm_tasks,
self.config, self.loop)
self.k8scluster = vim_sdn.K8sClusterLcm(self.db, self.msg, self.fs,
self.lcm_tasks, self.config,
self.loop)
self.k8srepo = vim_sdn.K8sRepoLcm(self.db, self.msg, self.fs,
self.lcm_tasks, self.config,
self.loop)
async def instantiate(self, nsir_id, nsilcmop_id):
# Try to lock HA task here
task_is_locked_by_me = self.lcm_tasks.lock_HA('nsi', 'nsilcmops',
nsilcmop_id)
if not task_is_locked_by_me:
return
logging_text = "Task netslice={} instantiate={} ".format(
nsir_id, nsilcmop_id)
self.logger.debug(logging_text + "Enter")
# get all needed from database
exc = None
db_nsir = None
db_nsilcmop = None
db_nsir_update = {"_admin.nsilcmop": nsilcmop_id}
db_nsilcmop_update = {}
nsilcmop_operation_state = None
vim_2_RO = {}
RO = ROclient.ROClient(self.loop, **self.ro_config)
def ip_profile_2_RO(ip_profile):
RO_ip_profile = deepcopy((ip_profile))
if "dns-server" in RO_ip_profile:
if isinstance(RO_ip_profile["dns-server"], list):
RO_ip_profile["dns-address"] = []
for ds in RO_ip_profile.pop("dns-server"):
RO_ip_profile["dns-address"].append(ds['address'])
else:
RO_ip_profile["dns-address"] = RO_ip_profile.pop(
"dns-server")
if RO_ip_profile.get("ip-version") == "ipv4":
RO_ip_profile["ip-version"] = "IPv4"
if RO_ip_profile.get("ip-version") == "ipv6":
RO_ip_profile["ip-version"] = "IPv6"
if "dhcp-params" in RO_ip_profile:
RO_ip_profile["dhcp"] = RO_ip_profile.pop("dhcp-params")
return RO_ip_profile
def vim_account_2_RO(vim_account):
"""
Translate a RO vim_account from OSM vim_account params
:param ns_params: OSM instantiate params
:return: The RO ns descriptor
"""
if vim_account in vim_2_RO:
return vim_2_RO[vim_account]
db_vim = self.db.get_one("vim_accounts", {"_id": vim_account})
if db_vim["_admin"]["operationalState"] != "ENABLED":
raise LcmException(
"VIM={} is not available. operationalState={}".format(
vim_account, db_vim["_admin"]["operationalState"]))
RO_vim_id = db_vim["_admin"]["deployed"]["RO"]
vim_2_RO[vim_account] = RO_vim_id
return RO_vim_id
async def netslice_scenario_create(self, vld_item, nsir_id, db_nsir,
db_nsir_admin, db_nsir_update):
"""
Create a network slice VLD through RO Scenario
:param vld_id The VLD id inside nsir to be created
:param nsir_id The nsir id
"""
ip_vld = None
mgmt_network = False
RO_vld_sites = []
vld_id = vld_item["id"]
netslice_vld = vld_item
# logging_text = "Task netslice={} instantiate_vld={} ".format(nsir_id, vld_id)
# self.logger.debug(logging_text + "Enter")
vld_shared = None
for shared_nsrs_item in get_iterable(vld_item, "shared-nsrs-list"):
_filter = {
"_id.ne": nsir_id,
"_admin.nsrs-detailed-list.ANYINDEX.nsrId":
shared_nsrs_item
}
shared_nsi = self.db.get_one("nsis",
_filter,
fail_on_empty=False,
fail_on_more=False)
if shared_nsi:
for vlds in get_iterable(shared_nsi["_admin"]["deployed"],
"RO"):
if vld_id == vlds["vld_id"]:
vld_shared = {
"instance_scenario_id":
vlds["netslice_scenario_id"],
"osm_id":
vld_id
}
break
break
# Creating netslice-vld at RO
RO_nsir = deep_get(db_nsir, ("_admin", "deployed", "RO"), [])
if vld_id in RO_nsir:
db_nsir_update["_admin.deployed.RO"] = RO_nsir
# If netslice-vld doesn't exists then create it
else:
# TODO: Check VDU type in all descriptors finding SRIOV / PT
# Updating network names and datacenters from instantiation parameters for each VLD
RO_ns_params = {}
RO_ns_params["name"] = netslice_vld["name"]
RO_ns_params["datacenter"] = vim_account_2_RO(
db_nsir["instantiation_parameters"]["vimAccountId"])
for instantiation_params_vld in get_iterable(
db_nsir["instantiation_parameters"], "netslice-vld"):
if instantiation_params_vld.get(
"name") == netslice_vld["name"]:
ip_vld = deepcopy(instantiation_params_vld)
if netslice_vld.get("mgmt-network"):
mgmt_network = True
# Creating scenario if vim-network-name / vim-network-id are present as instantiation parameter
# Use vim-network-id instantiation parameter
vim_network_option = None
if ip_vld:
if ip_vld.get("vim-network-id"):
vim_network_option = "vim-network-id"
elif ip_vld.get("vim-network-name"):
vim_network_option = "vim-network-name"
if ip_vld.get("ip-profile"):
populate_dict(
RO_ns_params,
("networks", netslice_vld["name"], "ip-profile"),
ip_profile_2_RO(ip_vld["ip-profile"]))
if vim_network_option:
if ip_vld.get(vim_network_option):
if isinstance(ip_vld.get(vim_network_option), list):
for vim_net_id in ip_vld.get(vim_network_option):
for vim_account, vim_net in vim_net_id.items():
RO_vld_sites.append({
"netmap-use":
vim_net,
"datacenter":
vim_account_2_RO(vim_account)
})
elif isinstance(ip_vld.get(vim_network_option), dict):
for vim_account, vim_net in ip_vld.get(
vim_network_option).items():
RO_vld_sites.append({
"netmap-use":
vim_net,
"datacenter":
vim_account_2_RO(vim_account)
})
else:
RO_vld_sites.append({
"netmap-use":
ip_vld[vim_network_option],
"datacenter":
vim_account_2_RO(netslice_vld["vimAccountId"])
})
# Use default netslice vim-network-name from template
else:
for nss_conn_point_ref in get_iterable(
netslice_vld, "nss-connection-point-ref"):
if nss_conn_point_ref.get("vimAccountId"):
if nss_conn_point_ref[
"vimAccountId"] != netslice_vld[
"vimAccountId"]:
RO_vld_sites.append({
"netmap-create":
None,
"datacenter":
vim_account_2_RO(
nss_conn_point_ref["vimAccountId"])
})
if vld_shared:
populate_dict(
RO_ns_params,
("networks", netslice_vld["name"], "use-network"),
vld_shared)
if RO_vld_sites:
populate_dict(RO_ns_params,
("networks", netslice_vld["name"], "sites"),
RO_vld_sites)
RO_ns_params["scenario"] = {
"nets": [{
"name": netslice_vld["name"],
"external": mgmt_network,
"type": "bridge"
}]
}
# self.logger.debug(logging_text + step)
desc = await RO.create("ns", descriptor=RO_ns_params)
db_nsir_update_RO = {}
db_nsir_update_RO["netslice_scenario_id"] = desc["uuid"]
db_nsir_update_RO["vld_id"] = RO_ns_params["name"]
db_nsir_update["_admin.deployed.RO"].append(db_nsir_update_RO)
def overwrite_nsd_params(self, db_nsir, nslcmop):
RO_list = []
vld_op_list = []
vld = None
nsr_id = nslcmop.get("nsInstanceId")
# Overwrite instantiation parameters in netslice runtime
if db_nsir.get("_admin"):
if db_nsir["_admin"].get("deployed"):
db_admin_deployed_nsir = db_nsir["_admin"].get("deployed")
if db_admin_deployed_nsir.get("RO"):
RO_list = db_admin_deployed_nsir["RO"]
for RO_item in RO_list:
for netslice_vld in get_iterable(db_nsir["_admin"],
"netslice-vld"):
# if is equal vld of _admin with vld of netslice-vld then go for the CPs
if RO_item.get("vld_id") == netslice_vld.get("id"):
# Search the cp of netslice-vld that match with nst:netslice-subnet
for nss_cp_item in get_iterable(
netslice_vld, "nss-connection-point-ref"):
# Search the netslice-subnet of nst that match
for nss in get_iterable(db_nsir["_admin"],
"netslice-subnet"):
# Compare nss-ref equal nss from nst
if nss_cp_item["nss-ref"] == nss["nss-id"]:
db_nsds = self.db.get_one(
"nsds", {"_id": nss["nsdId"]})
# Go for nsd, and search the CP that match with nst:CP to get vld-id-ref
for cp_nsd in db_nsds.get(
"connection-point", ()):
if cp_nsd["name"] == nss_cp_item[
"nsd-connection-point-ref"]:
if nslcmop.get("operationParams"):
if nslcmop[
"operationParams"].get(
"nsName"
) == nss["nsName"]:
vld_id = RO_item["vld_id"]
netslice_scenario_id = RO_item[
"netslice_scenario_id"]
nslcmop_vld = {}
nslcmop_vld["ns-net"] = {
vld_id:
netslice_scenario_id
}
nslcmop_vld[
"name"] = cp_nsd[
"vld-id-ref"]
for vld in get_iterable(
nslcmop[
"operationParams"],
"vld"):
if vld["name"] == cp_nsd[
"vld-id-ref"]:
nslcmop_vld.update(
vld)
vld_op_list.append(
nslcmop_vld)
nslcmop["operationParams"]["vld"] = vld_op_list
self.update_db_2("nslcmops", nslcmop["_id"],
{"operationParams.vld": vld_op_list})
return nsr_id, nslcmop
try:
# wait for any previous tasks in process
await self.lcm_tasks.waitfor_related_HA('nsi', 'nsilcmops',
nsilcmop_id)
step = "Getting nsir={} from db".format(nsir_id)
db_nsir = self.db.get_one("nsis", {"_id": nsir_id})
step = "Getting nsilcmop={} from db".format(nsilcmop_id)
db_nsilcmop = self.db.get_one("nsilcmops", {"_id": nsilcmop_id})
start_deploy = time()
nsi_params = db_nsilcmop.get("operationParams")
if nsi_params and nsi_params.get("timeout_nsi_deploy"):
timeout_nsi_deploy = nsi_params["timeout_nsi_deploy"]
else:
timeout_nsi_deploy = self.timeout.get("nsi_deploy",
self.timeout_nsi_deploy)
# Empty list to keep track of network service records status in the netslice
nsir_admin = db_nsir_admin = db_nsir.get("_admin")
step = "Creating slice operational-status init"
# Slice status Creating
db_nsir_update["detailed-status"] = "creating"
db_nsir_update["operational-status"] = "init"
db_nsir_update["_admin.nsiState"] = "INSTANTIATED"
step = "Instantiating netslice VLDs before NS instantiation"
# Creating netslice VLDs networking before NS instantiation
db_nsir_update["detailed-status"] = step
self.update_db_2("nsis", nsir_id, db_nsir_update)
db_nsir_update["_admin.deployed.RO"] = db_nsir_admin["deployed"][
"RO"]
for vld_item in get_iterable(nsir_admin, "netslice-vld"):
await netslice_scenario_create(self, vld_item, nsir_id,
db_nsir, db_nsir_admin,
db_nsir_update)
self.update_db_2("nsis", nsir_id, db_nsir_update)
step = "Instantiating netslice subnets"
db_nsir_update["detailed-status"] = step
self.update_db_2("nsis", nsir_id, db_nsir_update)
db_nsir = self.db.get_one("nsis", {"_id": nsir_id})
# Check status of the VLDs and wait for creation
# netslice_scenarios = db_nsir["_admin"]["deployed"]["RO"]
# db_nsir_update_RO = deepcopy(netslice_scenarios)
# for netslice_scenario in netslice_scenarios:
# await netslice_scenario_check(self, netslice_scenario["netslice_scenario_id"],
# nsir_id, db_nsir_update_RO)
# db_nsir_update["_admin.deployed.RO"] = db_nsir_update_RO
# self.update_db_2("nsis", nsir_id, db_nsir_update)
# Iterate over the network services operation ids to instantiate NSs
step = "Instantiating Netslice Subnets"
db_nsir = self.db.get_one("nsis", {"_id": nsir_id})
nslcmop_ids = db_nsilcmop["operationParams"].get("nslcmops_ids")
for nslcmop_id in nslcmop_ids:
nslcmop = self.db.get_one("nslcmops", {"_id": nslcmop_id})
# Overwriting netslice-vld vim-net-id to ns
nsr_id, nslcmop = overwrite_nsd_params(self, db_nsir, nslcmop)
step = "Launching ns={} instantiate={} task".format(
nsr_id, nslcmop_id)
task = asyncio.ensure_future(
self.ns.instantiate(nsr_id, nslcmop_id))
self.lcm_tasks.register("ns", nsr_id, nslcmop_id,
"ns_instantiate", task)
# Wait until Network Slice is ready
step = " Waiting nsi ready."
nsrs_detailed_list_old = None
self.logger.debug(logging_text + step)
# For HA, it is checked from database, as the ns operation may be managed by other LCM worker
while time() <= start_deploy + timeout_nsi_deploy:
# Check ns instantiation status
nsi_ready = True
nsir = self.db.get_one("nsis", {"_id": nsir_id})
nsrs_detailed_list = nsir["_admin"]["nsrs-detailed-list"]
nsrs_detailed_list_new = []
for nslcmop_item in nslcmop_ids:
nslcmop = self.db.get_one("nslcmops",
{"_id": nslcmop_item})
status = nslcmop.get("operationState")
# TODO: (future improvement) other possible status: ROLLING_BACK,ROLLED_BACK
for nss in nsrs_detailed_list:
if nss["nsrId"] == nslcmop["nsInstanceId"]:
nss.update({
"nsrId":
nslcmop["nsInstanceId"],
"status":
nslcmop["operationState"],
"detailed-status":
nslcmop.get("detailed-status"),
"instantiated":
True
})
nsrs_detailed_list_new.append(nss)
if status not in [
"COMPLETED", "PARTIALLY_COMPLETED", "FAILED",
"FAILED_TEMP"
]:
nsi_ready = False
if nsrs_detailed_list_new != nsrs_detailed_list_old:
nsrs_detailed_list_old = nsrs_detailed_list_new
self.update_db_2(
"nsis", nsir_id,
{"_admin.nsrs-detailed-list": nsrs_detailed_list_new})
if nsi_ready:
error_list = []
step = "Network Slice Instance instantiated"
for nss in nsrs_detailed_list:
if nss["status"] in ("FAILED", "FAILED_TEMP"):
error_list.append("NS {} {}: {}".format(
nss["nsrId"], nss["status"],
nss["detailed-status"]))
if error_list:
step = "instantiating"
raise LcmException("; ".join(error_list))
break
# TODO: future improvement due to synchronism -> await asyncio.wait(vca_task_list, timeout=300)
await asyncio.sleep(5, loop=self.loop)
else: # timeout_nsi_deploy reached:
raise LcmException("Timeout waiting nsi to be ready.")
db_nsir_update["operational-status"] = "running"
db_nsir_update["detailed-status"] = "done"
db_nsir_update["config-status"] = "configured"
db_nsilcmop_update[
"operationState"] = nsilcmop_operation_state = "COMPLETED"
db_nsilcmop_update["statusEnteredTime"] = time()
db_nsilcmop_update["detailed-status"] = "done"
return
except (LcmException, DbException) as e:
self.logger.error(logging_text +
"Exit Exception while '{}': {}".format(step, e))
exc = e
except asyncio.CancelledError:
self.logger.error(logging_text +
"Cancelled Exception while '{}'".format(step))
exc = "Operation was cancelled"
except Exception as e:
exc = traceback.format_exc()
self.logger.critical(logging_text +
"Exit Exception {} while '{}': {}".format(
type(e).__name__, step, e),
exc_info=True)
finally:
if exc:
if db_nsir:
db_nsir_update["detailed-status"] = "ERROR {}: {}".format(
step, exc)
db_nsir_update["operational-status"] = "failed"
db_nsir_update["config-status"] = "configured"
if db_nsilcmop:
db_nsilcmop_update[
"detailed-status"] = "FAILED {}: {}".format(step, exc)
db_nsilcmop_update[
"operationState"] = nsilcmop_operation_state = "FAILED"
db_nsilcmop_update["statusEnteredTime"] = time()
try:
if db_nsir:
db_nsir_update["_admin.nsilcmop"] = None
self.update_db_2("nsis", nsir_id, db_nsir_update)
if db_nsilcmop:
self.update_db_2("nsilcmops", nsilcmop_id,
db_nsilcmop_update)
except DbException as e:
self.logger.error(logging_text +
"Cannot update database: {}".format(e))
if nsilcmop_operation_state:
try:
await self.msg.aiowrite(
"nsi", "instantiated", {
"nsir_id": nsir_id,
"nsilcmop_id": nsilcmop_id,
"operationState": nsilcmop_operation_state
})
except Exception as e:
self.logger.error(
logging_text +
"kafka_write notification Exception {}".format(e))
self.logger.debug(logging_text + "Exit")
self.lcm_tasks.remove("nsi", nsir_id, nsilcmop_id,
"nsi_instantiate")
async def terminate(self, nsir_id, nsilcmop_id):
# Try to lock HA task here
task_is_locked_by_me = self.lcm_tasks.lock_HA('nsi', 'nsilcmops',
nsilcmop_id)
if not task_is_locked_by_me:
return
logging_text = "Task nsi={} terminate={} ".format(nsir_id, nsilcmop_id)
self.logger.debug(logging_text + "Enter")
exc = None
db_nsir = None
db_nsilcmop = None
db_nsir_update = {"_admin.nsilcmop": nsilcmop_id}
db_nsilcmop_update = {}
RO = ROclient.ROClient(self.loop, **self.ro_config)
nsir_deployed = None
failed_detail = [] # annotates all failed error messages
nsilcmop_operation_state = None
autoremove = False # autoremove after terminated
try:
# wait for any previous tasks in process
await self.lcm_tasks.waitfor_related_HA('nsi', 'nsilcmops',
nsilcmop_id)
step = "Getting nsir={} from db".format(nsir_id)
db_nsir = self.db.get_one("nsis", {"_id": nsir_id})
nsir_deployed = deepcopy(db_nsir["_admin"].get("deployed"))
step = "Getting nsilcmop={} from db".format(nsilcmop_id)
db_nsilcmop = self.db.get_one("nsilcmops", {"_id": nsilcmop_id})
# TODO: Check if makes sense check the nsiState=NOT_INSTANTIATED when terminate
# CASE: Instance was terminated but there is a second request to terminate the instance
if db_nsir["_admin"]["nsiState"] == "NOT_INSTANTIATED":
return
# Slice status Terminating
db_nsir_update["operational-status"] = "terminating"
db_nsir_update["config-status"] = "terminating"
db_nsir_update["detailed-status"] = "Terminating Netslice subnets"
self.update_db_2("nsis", nsir_id, db_nsir_update)
# Gets the list to keep track of network service records status in the netslice
nsrs_detailed_list = []
# Iterate over the network services operation ids to terminate NSs
# TODO: (future improvement) look another way check the tasks instead of keep asking
# -> https://docs.python.org/3/library/asyncio-task.html#waiting-primitives
# steps: declare ns_tasks, add task when terminate is called, await asyncio.wait(vca_task_list, timeout=300)
step = "Terminating Netslice Subnets"
nslcmop_ids = db_nsilcmop["operationParams"].get("nslcmops_ids")
nslcmop_new = []
for nslcmop_id in nslcmop_ids:
nslcmop = self.db.get_one("nslcmops", {"_id": nslcmop_id})
nsr_id = nslcmop["operationParams"].get("nsInstanceId")
nss_in_use = self.db.get_list(
"nsis", {
"_admin.netslice-vld.ANYINDEX.shared-nsrs-list":
nsr_id,
"operational-status": {
"$nin": ["terminated", "failed"]
}
})
if len(nss_in_use) < 2:
task = asyncio.ensure_future(
self.ns.terminate(nsr_id, nslcmop_id))
self.lcm_tasks.register("ns", nsr_id, nslcmop_id,
"ns_instantiate", task)
nslcmop_new.append(nslcmop_id)
else:
# Update shared nslcmop shared with active nsi
netsliceInstanceId = db_nsir["_id"]
for nsis_item in nss_in_use:
if db_nsir["_id"] != nsis_item["_id"]:
netsliceInstanceId = nsis_item["_id"]
break
self.db.set_one("nslcmops", {"_id": nslcmop_id}, {
"operationParams.netsliceInstanceId":
netsliceInstanceId
})
self.db.set_one("nsilcmops", {"_id": nsilcmop_id},
{"operationParams.nslcmops_ids": nslcmop_new})
# Wait until Network Slice is terminated
step = nsir_status_detailed = " Waiting nsi terminated. nsi_id={}".format(
nsir_id)
nsrs_detailed_list_old = None
self.logger.debug(logging_text + step)
termination_timeout = 2 * 3600 # Two hours
while termination_timeout > 0:
# Check ns termination status
nsi_ready = True
db_nsir = self.db.get_one("nsis", {"_id": nsir_id})
nsrs_detailed_list = db_nsir["_admin"].get(
"nsrs-detailed-list")
nsrs_detailed_list_new = []
for nslcmop_item in nslcmop_ids:
nslcmop = self.db.get_one("nslcmops",
{"_id": nslcmop_item})
status = nslcmop["operationState"]
# TODO: (future improvement) other possible status: ROLLING_BACK,ROLLED_BACK
for nss in nsrs_detailed_list:
if nss["nsrId"] == nslcmop["nsInstanceId"]:
nss.update({
"nsrId":
nslcmop["nsInstanceId"],
"status":
nslcmop["operationState"],
"detailed-status":
nsir_status_detailed +
"; {}".format(nslcmop.get("detailed-status"))
})
nsrs_detailed_list_new.append(nss)
if status not in [
"COMPLETED", "PARTIALLY_COMPLETED", "FAILED",
"FAILED_TEMP"
]:
nsi_ready = False
if nsrs_detailed_list_new != nsrs_detailed_list_old:
nsrs_detailed_list_old = nsrs_detailed_list_new
self.update_db_2(
"nsis", nsir_id,
{"_admin.nsrs-detailed-list": nsrs_detailed_list_new})
if nsi_ready:
# Check if it is the last used nss and mark isinstantiate: False
db_nsir = self.db.get_one("nsis", {"_id": nsir_id})
nsrs_detailed_list = db_nsir["_admin"].get(
"nsrs-detailed-list")
for nss in nsrs_detailed_list:
_filter = {
"_admin.nsrs-detailed-list.ANYINDEX.nsrId":
nss["nsrId"],
"operational-status.ne":
"terminated",
"_id.ne":
nsir_id
}
nsis_list = self.db.get_one("nsis",
_filter,
fail_on_empty=False,
fail_on_more=False)
if not nsis_list:
nss.update({"instantiated": False})
step = "Network Slice Instance is terminated. nsi_id={}".format(
nsir_id)
for items in nsrs_detailed_list:
if "FAILED" in items.values():
raise LcmException(
"Error terminating NSI: {}".format(nsir_id))
break
await asyncio.sleep(5, loop=self.loop)
termination_timeout -= 5
if termination_timeout <= 0:
raise LcmException(
"Timeout waiting nsi to be terminated. nsi_id={}".format(
nsir_id))
# Delete netslice-vlds
RO_nsir_id = RO_delete_action = None
for nsir_deployed_RO in get_iterable(nsir_deployed, "RO"):
RO_nsir_id = nsir_deployed_RO.get("netslice_scenario_id")
try:
step = db_nsir_update[
"detailed-status"] = "Deleting netslice-vld at RO"
db_nsilcmop_update[
"detailed-status"] = "Deleting netslice-vld at RO"
self.logger.debug(logging_text + step)
desc = await RO.delete("ns", RO_nsir_id)
RO_delete_action = desc["action_id"]
nsir_deployed_RO["vld_delete_action_id"] = RO_delete_action
nsir_deployed_RO["vld_status"] = "DELETING"
db_nsir_update["_admin.deployed"] = nsir_deployed
self.update_db_2("nsis", nsir_id, db_nsir_update)
if RO_delete_action:
# wait until NS is deleted from VIM
step = "Waiting ns deleted from VIM. RO_id={}".format(
RO_nsir_id)
self.logger.debug(logging_text + step)
except ROclient.ROClientException as e:
if e.http_code == 404: # not found
nsir_deployed_RO["vld_id"] = None
nsir_deployed_RO["vld_status"] = "DELETED"
self.logger.debug(
logging_text +
"RO_ns_id={} already deleted".format(RO_nsir_id))
elif e.http_code == 409: # conflict
failed_detail.append(
"RO_ns_id={} delete conflict: {}".format(
RO_nsir_id, e))
self.logger.debug(logging_text + failed_detail[-1])
else:
failed_detail.append(
"RO_ns_id={} delete error: {}".format(
RO_nsir_id, e))
self.logger.error(logging_text + failed_detail[-1])
if failed_detail:
self.logger.error(logging_text + " ;".join(failed_detail))
db_nsir_update["operational-status"] = "failed"
db_nsir_update[
"detailed-status"] = "Deletion errors " + "; ".join(
failed_detail)
db_nsilcmop_update["detailed-status"] = "; ".join(
failed_detail)
db_nsilcmop_update[
"operationState"] = nsilcmop_operation_state = "FAILED"
db_nsilcmop_update["statusEnteredTime"] = time()
else:
db_nsir_update["operational-status"] = "terminating"
db_nsir_update["config-status"] = "terminating"
db_nsir_update["_admin.nsiState"] = "NOT_INSTANTIATED"
db_nsilcmop_update[
"operationState"] = nsilcmop_operation_state = "COMPLETED"
db_nsilcmop_update["statusEnteredTime"] = time()
if db_nsilcmop["operationParams"].get("autoremove"):
autoremove = True
db_nsir_update["detailed-status"] = "done"
db_nsir_update["operational-status"] = "terminated"
db_nsir_update["config-status"] = "terminated"
db_nsilcmop_update["statusEnteredTime"] = time()
db_nsilcmop_update["detailed-status"] = "done"
return
except (LcmException, DbException) as e:
self.logger.error(logging_text +
"Exit Exception while '{}': {}".format(step, e))
exc = e
except asyncio.CancelledError:
self.logger.error(logging_text +
"Cancelled Exception while '{}'".format(step))
exc = "Operation was cancelled"
except Exception as e:
exc = traceback.format_exc()
self.logger.critical(logging_text +
"Exit Exception {} while '{}': {}".format(
type(e).__name__, step, e),
exc_info=True)
finally:
if exc:
if db_nsir:
db_nsir_update["_admin.deployed"] = nsir_deployed
db_nsir_update["detailed-status"] = "ERROR {}: {}".format(
step, exc)
db_nsir_update["operational-status"] = "failed"
if db_nsilcmop:
db_nsilcmop_update[
"detailed-status"] = "FAILED {}: {}".format(step, exc)
db_nsilcmop_update[
"operationState"] = nsilcmop_operation_state = "FAILED"
db_nsilcmop_update["statusEnteredTime"] = time()
try:
if db_nsir:
db_nsir_update["_admin.deployed"] = nsir_deployed
db_nsir_update["_admin.nsilcmop"] = None
self.update_db_2("nsis", nsir_id, db_nsir_update)
if db_nsilcmop:
self.update_db_2("nsilcmops", nsilcmop_id,
db_nsilcmop_update)
except DbException as e:
self.logger.error(logging_text +
"Cannot update database: {}".format(e))
if nsilcmop_operation_state:
try:
await self.msg.aiowrite(
"nsi",
"terminated", {
"nsir_id": nsir_id,
"nsilcmop_id": nsilcmop_id,
"operationState": nsilcmop_operation_state,
"autoremove": autoremove
},
loop=self.loop)
except Exception as e:
self.logger.error(
logging_text +
"kafka_write notification Exception {}".format(e))
self.logger.debug(logging_text + "Exit")
self.lcm_tasks.remove("nsi", nsir_id, nsilcmop_id, "nsi_terminate")
async def update(self,
add_jobs: dict = None,
remove_jobs: list = None) -> bool:
"""
:param add_jobs: dictionary with {job_id_1: job_content, job_id_2: job_content}
:param remove_jobs: list with jobs to remove [job_id_1, job_id_2]
:return: result. If false prometheus denies this configuration. Exception on error
"""
for retry in range(4):
result = True
if retry: # first time do not wait
await asyncio.sleep(self.PROMETHEUS_LOCKED_TIME / 2,
loop=self.loop)
# lock database
now = time()
if not self.db.set_one("admin",
q_filter={
"_id":
"prometheus",
"_admin.locked_at.lt":
now - self.PROMETHEUS_LOCKED_TIME
},
update_dict={
"_admin.locked_at": now,
"_admin.locked_by": self.worker_id
},
fail_on_empty=False):
continue
# read database
prometheus_data = self.db.get_one("admin", {"_id": "prometheus"})
update_dict = {"_admin.locked_at": 0, "_admin.locked_by": None}
# Make changes from prometheus_incremental
push_dict = pull_dict = None
if add_jobs or remove_jobs:
log_text_list = []
if add_jobs:
log_text_list.append("adding jobs: {}".format(
list(add_jobs.keys())))
prometheus_data["scrape_configs"].update(add_jobs)
push_dict = {
"scrape_configs." + job_id: job_data
for job_id, job_data in add_jobs.items()
}
elif remove_jobs:
log_text_list.append("removing jobs: {}".format(
list(remove_jobs)))
for job_id in remove_jobs:
prometheus_data["scrape_configs"].pop(job_id, None)
pull_dict = {
"scrape_configs." + job_id: None
for job_id in remove_jobs
}
self.logger.debug("Updating. " + ". ".join(log_text_list))
if not await self.send_data(prometheus_data):
self.logger.error(
"Cannot update add_jobs: {}. remove_jobs: {}".format(
add_jobs, remove_jobs))
push_dict = pull_dict = None
result = False
# unblock database
if push_dict:
update_dict.update(push_dict)
if push_dict or pull_dict:
update_dict["_admin.modified_at"] = now
if not self.db.set_one("admin", {
"_id": "prometheus",
"_admin.locked_at": now,
"_admin.locked_by": self.worker_id
},
update_dict=update_dict,
unset=pull_dict,
fail_on_empty=False):
continue
return result
raise LcmException(
"Cannot update prometheus database. Reached max retries")