def handle(self):
cdp = CiscoCDPMib(self.agent)
stampcheck = yield self._get_stampcheck(cdp)
remote_table_has_changed = yield stampcheck.is_changed()
need_to_collect = remote_table_has_changed
if not remote_table_has_changed:
cache = yield self._get_cached_neighbors()
if cache is not None:
self._logger.debug("Using cached CDP neighbors")
self.neighbors = cache
else:
self._logger.debug(
"CDP cache table didn't change, but local cache was empty")
need_to_collect = True
if need_to_collect:
self._logger.debug("collecting CDP cache table")
self.neighbors = yield cdp.get_cdp_neighbors()
if self.neighbors:
self._logger.debug("CDP neighbors:\n %r", self.neighbors)
yield run_in_thread(self._process_neighbors)
yield self._save_cached_neighbors(self.neighbors)
else:
self._logger.debug("No CDP neighbors to process")
# Store sentinels to signal that CDP neighbors have been processed
shadows.AdjacencyCandidate.sentinel(self.containers, SOURCE)
shadows.UnrecognizedNeighbor.sentinel(self.containers, SOURCE)
stampcheck.save()
def can_handle(cls, netbox):
base_can_handle = yield defer.maybeDeferred(
super(StatSensors, cls).can_handle, netbox)
if base_can_handle:
i_can_handle = yield run_in_thread(cls._has_sensors, netbox)
defer.returnValue(i_can_handle)
defer.returnValue(base_can_handle)
def _get_cached_neighbors(self):
"""Retrieves a cached version of the remote neighbor table"""
value = yield run_in_thread(
manage.NetboxInfo.cache_get,
self.netbox,
INFO_KEY_NAME,
INFO_VAR_NEIGHBORS_CACHE
)
defer.returnValue(value)
def _save_cached_neighbors(self, neighbors):
"""Saves a cached a copy of the remote neighbor table"""
yield run_in_thread(
manage.NetboxInfo.cache_set,
self.netbox,
INFO_KEY_NAME,
INFO_VAR_NEIGHBORS_CACHE,
neighbors
)
def _get_cached_remote_table(self):
"""Retrieves a cached version of the remote neighbor table"""
value = yield run_in_thread(
manage.NetboxInfo.cache_get,
self.netbox,
INFO_KEY_LLDP_INFO,
INFO_VAR_REMOTES_CACHE,
)
defer.returnValue(value)
def _save_cached_remote_table(self, remote_table):
"""Saves a cached copy of the remote neighbor table"""
yield run_in_thread(
manage.NetboxInfo.cache_set,
self.netbox,
INFO_KEY_LLDP_INFO,
INFO_VAR_REMOTES_CACHE,
remote_table,
)
def _get_type_from_db(self):
"""Loads from db a type object matching the sysobjectid."""
def _single_result(result):
if result:
return result[0]
# Look up existing type entry
types = manage.NetboxType.objects.filter(sysobjectid=self.sysobjectid)
df = db.run_in_thread(storage.shadowify_queryset_and_commit, types)
df.addCallback(_single_result)
return df
def handle(self):
self._logger.debug("snmp version from db: %s",
self.netbox.snmp_version)
was_down = yield db.run_in_thread(self._currently_down)
is_ok = yield self._do_check()
if is_ok and was_down:
yield self._mark_as_up()
elif not is_ok and not was_down:
yield self._mark_as_down()
raise SuggestedReschedule(delay=60)
def handle(self):
self._logger.debug("snmp version from db: %s", self.netbox.snmp_version)
was_down = yield db.run_in_thread(self._currently_down)
is_ok = yield self._do_check()
if is_ok and was_down:
yield self._mark_as_up()
elif not is_ok:
# Always send down events; eventengine will ignore any duplicates
yield self._mark_as_down()
raise SuggestedReschedule(delay=60)
def handle(self):
mib = LLDPMib(self.agent)
stampcheck = yield self._stampcheck(mib)
need_to_collect = yield stampcheck.is_changed()
if need_to_collect:
self.remote = yield mib.get_remote_table()
if self.remote:
self._logger.debug("LLDP neighbors:\n %s", pformat(self.remote))
yield run_in_thread(self._process_remote)
stampcheck.save()
def handle(self):
cdp = CiscoCDPMib(self.agent)
stampcheck = yield self._stampcheck(cdp)
need_to_collect = yield stampcheck.is_changed()
if need_to_collect:
cache = yield cdp.get_cdp_neighbors()
if cache:
self._logger.debug("found CDP cache data: %r", cache)
self.cache = cache
yield run_in_thread(self._process_cache)
stampcheck.save()
def handle(self):
mib = LLDPMib(self.agent)
stampcheck = yield self._stampcheck(mib)
need_to_collect = yield stampcheck.is_changed()
if need_to_collect:
self.remote = yield mib.get_remote_table()
if self.remote:
self._logger.debug("LLDP neighbors:\n %s",
pformat(self.remote))
yield run_in_thread(self._process_remote)
stampcheck.save()
def handle(self):
if self.netbox.master:
defer.returnValue(None)
netboxes = yield db.run_in_thread(self._get_netbox_list)
bandwidth = yield self._collect_bandwidth(netboxes)
cpu = yield self._collect_cpu(netboxes)
sysuptime = yield self._collect_sysuptime(netboxes)
memory = yield self._collect_memory(netboxes)
metrics = bandwidth + cpu + sysuptime + memory
if metrics:
send_metrics(metrics)
def _get_type_from_db(self):
"""Loads from db a type object matching the sysobjectid."""
def _single_result(result):
if result:
return result[0]
# Look up existing type entry
types = manage.NetboxType.objects.filter(sysobjectid=self.sysobjectid)
df = db.run_in_thread(storage.shadowify_queryset_and_commit,
types)
df.addCallback(_single_result)
return df
def handle(self):
units = yield db.run_in_thread(self._get_database_unit_list)
state_map = {}
for unit in units:
old_state = unit.up
new_state = yield self._retrieve_current_unit_state(unit)
state_map[unit] = new_state
if old_state != new_state:
yield self._handle_state_change(unit, new_state)
defer.returnValue(True)
def handle(self):
cdp = CiscoCDPMib(self.agent)
stampcheck = yield self._stampcheck(cdp)
need_to_collect = yield stampcheck.is_changed()
if need_to_collect:
cache = yield cdp.get_cdp_neighbors()
if cache:
self._logger.debug("found CDP cache data: %r", cache)
self.cache = cache
yield run_in_thread(self._process_cache)
stampcheck.save()
def handle(self):
sensors = yield run_in_thread(self._get_sensors)
self._logger.debug("retrieving data from %d sensors", len(sensors))
oids = sensors.keys()
requests = [
oids[x:x + MAX_SENSORS_PER_REQUEST]
for x in range(0, len(oids), MAX_SENSORS_PER_REQUEST)
]
for req in requests:
data = yield self.agent.get(req).addCallback(
self._response_to_metrics, sensors)
self._logger.debug("got data from sensors: %r", data)
def handle(self):
if self.netbox.master:
yield self._log_instance_details()
defer.returnValue(None)
timestamp = time.time()
stats = yield self._get_stats()
netboxes = yield db.run_in_thread(self._get_netbox_list)
tuples = list(self._make_metrics(stats, netboxes=netboxes, timestamp=timestamp))
if tuples:
self._logger.debug("Counters collected")
send_metrics(tuples)
def _log_instance_details(self):
def _get_master_and_instance_list():
netbox = manage.Netbox.objects.get(id=self.netbox.id)
my_ifcs = netbox.interface_set.values_list('ifname', flat=True)
masters_ifcs = netbox.master.interface_set.values_list(
'ifname', flat=True)
local_ifcs = set(masters_ifcs) - set(my_ifcs)
return netbox.master.sysname, local_ifcs
if self._logger.isEnabledFor(logging.DEBUG):
master, ifcs = yield db.run_in_thread(_get_master_and_instance_list)
self._logger.debug("local interfaces (that do not exist on master "
"%s): %r", master, ifcs)
def load(self):
"Loads existing timestamps from db"
def _unpickle():
try:
info = manage.NetboxInfo.objects.get(
netbox__id=self._get_netbox().id,
key=INFO_KEY_NAME, variable=self.var_name)
except manage.NetboxInfo.DoesNotExist:
return None
try:
return pickle.loads(str(info.value))
except Exception:
return None
self.loaded_times = yield db.run_in_thread(_unpickle)
defer.returnValue(self.loaded_times)
def create_new_type(self):
"""Creates a new NetboxType from the collected sysObjectID."""
vendor_id = yield db.run_in_thread(get_vendor_id, self.sysobjectid)
vendor = self.containers.factory(vendor_id, shadows.Vendor)
vendor.id = vendor_id
type_ = self.containers.factory(self.sysobjectid, shadows.NetboxType)
type_.vendor = vendor
type_.name = self.sysobjectid
type_.sysobjectid = self.sysobjectid
def _set_sysdescr(descr):
self._logger.debug("Creating new type with descr=%r", descr)
type_.description = descr
return type_
yield self.snmpv2_mib.get_sysDescr().addCallback(_set_sysdescr)
def on_netbox_type_changed(self, netbox_id, new_type, **kwargs):
"""Performs various cleanup and reload actions on a netbox type change
signal.
The netbox' data are cleaned up, and the next netbox data reload is
scheduled to take place immediately.
"""
sysname = netbox_id in self.netboxes and \
self.netboxes[netbox_id].sysname or str(netbox_id)
self._logger.info("Cancelling all jobs for %s due to type change.",
sysname)
self.cancel_netbox_scheduler(netbox_id)
df = db.run_in_thread(shadows.Netbox.cleanup_replaced_netbox,
netbox_id, new_type)
return df.addCallback(lambda x: self._start_netbox_reload_loop())
def on_netbox_type_changed(self, netbox_id, new_type, **_kwargs):
"""Performs various cleanup and reload actions on a netbox type change
signal.
The netbox' data are cleaned up, and the next netbox data reload is
scheduled to take place immediately.
"""
sysname = netbox_id in self.netboxes and \
self.netboxes[netbox_id].sysname or str(netbox_id)
self._logger.info("Cancelling all jobs for %s due to type change.",
sysname)
self.cancel_netbox_scheduler(netbox_id)
df = db.run_in_thread(shadows.Netbox.cleanup_replaced_netbox,
netbox_id, new_type)
return df.addCallback(lambda x: self._start_netbox_reload_loop())
def _create_new_type(self, oid):
"""Creates a new NetboxType from the given sysobjectid."""
self._logger.debug("Creating a new type from %r", oid)
description = yield self.snmpv2_mib.get_sysDescr()
def _create():
vendor = self._get_vendor(oid)
type_ = manage.NetboxType(
vendor=vendor,
name=str(oid),
sysobjectid=str(oid).strip("."),
description=description,
)
type_.save()
return type_
new_type = yield db.run_in_thread(_create)
defer.returnValue(new_type)
def handle(self):
mib = LLDPMib(self.agent)
stampcheck = yield self._stampcheck(mib)
need_to_collect = yield stampcheck.is_changed()
if need_to_collect:
self._logger.debug("collecting LLDP remote table")
self.remote = yield mib.get_remote_table()
if self.remote:
self._logger.debug("LLDP neighbors:\n %s",
pformat(self.remote))
yield run_in_thread(self._process_remote)
# Store sentinel to signal that LLDP neighbors have been processed
shadows.AdjacencyCandidate.sentinel(self.containers, SOURCE)
else:
self._logger.debug("LLDP remote table seems unchanged")
stampcheck.save()
def _load_existing_mappings(self):
"""Load the existing ARP records for this box from the db.
Returns:
A deferred whose result is a dictionary: { (ip, mac): arpid }
"""
self._logger.debug("Loading open arp records from database")
open_arp_records_queryset = manage.Arp.objects.filter(
netbox__id=self.netbox.id,
end_time__gte=datetime.max).values('id', 'ip', 'mac')
open_arp_records = yield db.run_in_thread(
storage.shadowify_queryset_and_commit, open_arp_records_queryset)
self._logger.debug("Loaded %d open records from arp",
len(open_arp_records))
open_mappings = dict(((IP(arp['ip']), arp['mac']), arp['id'])
for arp in open_arp_records)
defer.returnValue(open_mappings)
def handle(self):
cdp = CiscoCDPMib(self.agent)
stampcheck = yield self._stampcheck(cdp)
need_to_collect = yield stampcheck.is_changed()
if need_to_collect:
self._logger.debug("collecting CDP cache table")
cache = yield cdp.get_cdp_neighbors()
if cache:
self._logger.debug("found CDP cache data: %r", cache)
self.cache = cache
yield run_in_thread(self._process_cache)
# Store sentinel to signal that CDP neighbors have been processed
shadows.AdjacencyCandidate.sentinel(self.containers, SOURCE)
else:
self._logger.debug("CDP cache table seems unchanged")
stampcheck.save()
def _load_existing_mappings(self):
"""Load the existing ARP records for this box from the db.
Returns:
A deferred whose result is a dictionary: { (ip, mac): arpid }
"""
self._logger.debug("Loading open arp records from database")
open_arp_records_queryset = manage.Arp.objects.filter(
netbox__id=self.netbox.id,
end_time__gte=datetime.max).values('id', 'ip', 'mac')
open_arp_records = yield db.run_in_thread(
storage.shadowify_queryset_and_commit,
open_arp_records_queryset)
self._logger.debug("Loaded %d open records from arp",
len(open_arp_records))
open_mappings = dict(((IP(arp['ip']), arp['mac']), arp['id'])
for arp in open_arp_records)
defer.returnValue(open_mappings)
def _log_job_externally(self, success=True):
"""Logs a job to the database"""
duration = self.get_current_runtime()
duration_in_seconds = (
duration.days * 86400 + duration.seconds + duration.microseconds / 1e6
)
timestamp = time.time()
def _create_record(timestamp):
netbox = manage.Netbox.objects.get(id=self.netbox.id)
if netbox.deleted_at:
_logger.info(
"Not logging job to db; delete of this IP device"
" was requested at %s",
netbox.deleted_at,
)
return
log = manage.IpdevpollJobLog(
netbox_id=self.netbox.id,
job_name=self.name,
end_time=datetime.datetime.fromtimestamp(timestamp),
duration=duration_in_seconds,
success=success,
interval=self.interval,
)
log.save()
def _log_to_graphite():
prefix = metric_prefix_for_ipdevpoll_job(self.netbox.sysname, self.name)
runtime_path = prefix + ".runtime"
runtime = (runtime_path, (timestamp, duration_in_seconds))
send_metrics([runtime])
_log_to_graphite()
try:
yield db.run_in_thread(_create_record, timestamp)
except db.ResetDBConnectionError:
pass # this is being logged all over the place at the moment
except Exception as error:
_logger.warning("failed to log job to database: %s", error)
def _save_container(self):
"""
Parses the container and finds a sane storage order. We do this
so we get ForeignKeys stored before the objects that are using them
are stored.
"""
@db.cleanup_django_debug_after
def complete_save_cycle():
# Traverse all the classes in the container repository and
# generate the storage queue
self._populate_storage_queue()
# Prepare all shadow objects for storage.
self._prepare_containers_for_save()
# Actually save to the database
result = self._perform_save()
self._log_timed_result(result, "Storing to database complete")
# Do cleanup for the known container classes.
self._cleanup_containers_after_save()
df = db.run_in_thread(complete_save_cycle)
return df
def save_container(self):
"""
Parses the container and finds a sane storage order. We do this
so we get ForeignKeys stored before the objects that are using them
are stored.
"""
@db.autocommit
@db.cleanup_django_debug_after
def complete_save_cycle():
# Traverse all the classes in the container repository and
# generate the storage queue
self.populate_storage_queue()
# Prepare all shadow objects for storage.
self.prepare_containers_for_save()
# Actually save to the database
result = self.perform_save()
self.log_timed_result(result, "Storing to database complete")
# Do cleanup for the known container classes.
self.cleanup_containers_after_save()
df = db.run_in_thread(complete_save_cycle)
return df
def handle(self):
"""Gets forwarding tables from Q-BRIDGE-MIB. It that fails, reverts to
BRIDGE-MIB, with optional community string indexing on Cisco.
"""
fdb = yield self._get_dot1q_mac_port_mapping()
self._log_fdb_stats("Q-BRIDGE-MIB", fdb)
if not fdb:
fdb = yield self._get_dot1d_mac_port_mapping()
self._log_fdb_stats("BRIDGE-MIB", fdb)
self.fdb = fdb
self.monitored = yield db.run_in_thread(get_netbox_macs)
self.my_macs = set(mac for mac, netboxid in self.monitored.items()
if netboxid == self.netbox.id)
self._classify_ports()
self._store_cam_records()
self._store_adjacency_candidates()
self.blocking = yield self._get_dot1d_stp_blocking()
def log_job_externally(job_handler, success=True, interval=None):
"""Logs a job to the database"""
duration = job_handler.get_current_runtime()
duration_in_seconds = (duration.days * 86400 +
duration.seconds +
duration.microseconds / 1e6)
timestamp = time.time()
@db.autocommit
def _create_record(timestamp):
log = manage.IpdevpollJobLog(
netbox_id=job_handler.netbox.id,
job_name=job_handler.name,
end_time=datetime.datetime.fromtimestamp(timestamp),
duration=duration_in_seconds,
success=success,
interval=interval
)
log.save()
def _log_to_graphite():
prefix = metric_prefix_for_ipdevpoll_job(job_handler.netbox.sysname,
job_handler.name)
runtime_path = prefix + ".runtime"
runtime = (runtime_path, (timestamp, duration_in_seconds))
send_metrics([runtime])
counter_path = (
prefix + (".success-count" if success else ".failure-count"))
_COUNTERS.increment(counter_path)
_COUNTERS.start()
_log_to_graphite()
try:
yield db.run_in_thread(_create_record, timestamp)
except Exception, error:
_logger.warning("failed to log job to database: %s", error)
def _update_prefix_cache(cls):
cls.prefix_cache_update_time = datetime.now()
df = db.run_in_thread(cls._load_prefixes_synchronously)
df.addCallback(cls._update_prefix_cache_with_result)
return df
def load_all(self):
"""Asynchronously load netboxes from database."""
return run_in_thread(self.load_all_s)
def _mark_as_up(self):
self._logger.warning("SNMP agent up again on %s", self.netbox.sysname)
yield db.run_in_thread(self._dispatch_up_event)
def _mark_as_down(self):
self._logger.warning("SNMP agent down on %s", self.netbox.sysname)
yield db.run_in_thread(self._dispatch_down_event)
def _mark_as_down(self):
if self.netbox.id not in self.down_set:
self._logger.warning("SNMP agent down on %s", self.netbox.sysname)
self.down_set.add(self.netbox.id)
yield db.run_in_thread(self._dispatch_down_event)
def _handle_state_change(self, unit, new_state):
self._logger.info("%s state changed from %s to %s", unit.name, unit.up,
new_state)
yield db.run_in_thread(self._update_internal_state, unit, new_state)
yield db.run_in_thread(self._post_event, unit, new_state)
def can_handle(cls, netbox):
daddy_says_ok = super(CDP, cls).can_handle(netbox)
has_ifcs = yield run_in_thread(cls._has_interfaces, netbox)
defer.returnValue(has_ifcs and daddy_says_ok)
def _mark_as_up(self):
if self.netbox.id in self.down_set:
self._logger.warning("SNMP agent up again on %s",
self.netbox.sysname)
self.down_set.remove(self.netbox.id)
yield db.run_in_thread(self._dispatch_up_event)
def can_handle(cls, netbox):
daddy_says_ok = super(CDP, cls).can_handle(netbox)
has_ifcs = yield run_in_thread(cls._has_interfaces, netbox)
defer.returnValue(has_ifcs and daddy_says_ok)