def MarkDone(self, responses):
"""Mark a client as done."""
client_id = responses.request.client_id
# Open child flow and account its' reported resource usage
flow_path = responses.status.child_session_id
status = responses.status
resources = rdfvalue.ClientResources()
resources.client_id = client_id
resources.session_id = flow_path
resources.cpu_usage.user_cpu_time = status.cpu_time_used.user_cpu_time
resources.cpu_usage.system_cpu_time = status.cpu_time_used.system_cpu_time
resources.network_bytes_sent = status.network_bytes_sent
self.state.context.usage_stats.RegisterResources(resources)
if responses.success:
msg = "Flow %s completed." % self.state.context.description,
self.LogResult(client_id, msg)
with self.lock:
self.processed_responses = True
msgs = [
rdfvalue.GrrMessage(payload=response, source=client_id)
for response in responses
]
self.state.context.results_collection.AddAll(msgs)
else:
self.LogClientError(client_id,
log_message=utils.SmartStr(responses.status))
self.MarkClientDone(client_id)
def InitializeContext(self, args):
"""Initializes the context of this flow."""
if args is None:
args = rdfvalue.FlowRunnerArgs()
context = flows.DataObject(
args=args,
backtrace=None,
client_resources=rdfvalue.ClientResources(),
create_time=rdfvalue.RDFDatetime().Now(),
creator=self.token.username,
current_state="Start",
network_bytes_sent=0,
next_outbound_id=1,
next_processed_request=1,
next_states=set(),
output=self._CreateOutputCollection(args),
outstanding_requests=0,
remaining_cpu_quota=args.cpu_limit,
state=rdfvalue.Flow.State.RUNNING,
user=self.token.username,
# Have we sent a notification to the user.
user_notified=False,
)
# Store the context in the flow_obj for next time.
self.flow_obj.state.Register("context", context)
return context
def StoreResourceUsage(self, responses, client_id):
"""Open child flow and account its' reported resource usage."""
flow_path = responses.status.child_session_id
status = responses.status
resources = rdfvalue.ClientResources()
resources.client_id = client_id
resources.session_id = flow_path
resources.cpu_usage.user_cpu_time = status.cpu_time_used.user_cpu_time
resources.cpu_usage.system_cpu_time = status.cpu_time_used.system_cpu_time
resources.network_bytes_sent = status.network_bytes_sent
self.state.context.usage_stats.RegisterResources(resources)
def InitializeContext(self, args):
"""Initializes the context of this flow."""
if args is None:
args = rdfvalue.FlowRunnerArgs()
output_collection = self._CreateOutputCollection(args)
# Output collection is nullified when flow is terminated, so we're
# keeping the urn separately for further reference.
output_urn = (output_collection is not None) and output_collection.urn
output_plugins_states = []
for plugin_descriptor in args.output_plugins:
plugin_class = plugin_descriptor.GetPluginClass()
plugin = plugin_class(output_urn,
args=plugin_descriptor.plugin_args,
token=self.token)
try:
plugin.Initialize()
output_plugins_states.append((plugin_descriptor, plugin.state))
except Exception as e: # pylint: disable=broad-except
self.Log("Plugin %s failed to initialize (%s), ignoring it." %
(plugin, e))
context = utils.DataObject(
args=args,
backtrace=None,
client_resources=rdfvalue.ClientResources(),
create_time=rdfvalue.RDFDatetime().Now(),
creator=args.creator or self.token.username,
current_state="Start",
# If not None, kill-stuck-flow notification is scheduled at the given
# time.
kill_timestamp=None,
network_bytes_sent=0,
next_outbound_id=1,
next_processed_request=1,
next_states=set(),
output=output_collection,
output_plugins_states=output_plugins_states,
output_urn=output_urn,
outstanding_requests=0,
remaining_cpu_quota=args.cpu_limit,
state=rdfvalue.Flow.State.RUNNING,
# Have we sent a notification to the user.
user_notified=False,
)
# Store the context in the flow_obj for next time.
self.flow_obj.state.Register("context", context)
return context
def InitializeContext(self, args):
"""Initializes the context of this hunt."""
if args is None:
args = HuntRunnerArgs()
# For large hunts, checking client limits creates a high load on the foreman
# since it needs to read the hunt object's client list. We therefore don't
# allow setting it for large hunts. Note that client_limit of 0 means
# unlimited which is allowed (the foreman then does not need to check the
# client list)..
if args.client_limit > 1000:
raise RuntimeError("Please specify client_limit <= 1000.")
context = utils.DataObject(
args=args,
backtrace=None,
client_resources=rdfvalue.ClientResources(),
create_time=rdfvalue.RDFDatetime().Now(),
creator=self.token.username,
expires=rdfvalue.RDFDatetime().Now(),
# If not None, kill-stuck-flow notification is scheduled at the given
# time.
kill_timestamp=None,
network_bytes_sent=0,
next_client_due=0,
next_outbound_id=1,
next_processed_request=1,
next_states=set(),
outstanding_requests=0,
current_state=None,
start_time=rdfvalue.RDFDatetime().Now(),
# Hunts are always in the running state.
state=rdfvalue.Flow.State.RUNNING,
usage_stats=rdfvalue.ClientResourcesStats(),
remaining_cpu_quota=args.cpu_limit,
)
# Store the context in the flow_obj for next time.
self.flow_obj.state.Register("context", context)
return context
def InitializeContext(self, args):
"""Initializes the context of this flow."""
if args is None:
args = rdfvalue.FlowRunnerArgs()
output_collection = self._CreateOutputCollection(args)
context = utils.DataObject(
args=args,
backtrace=None,
client_resources=rdfvalue.ClientResources(),
create_time=rdfvalue.RDFDatetime().Now(),
creator=args.creator or self.token.username,
current_state="Start",
# If not None, kill-stuck-flow notification is scheduled at the given
# time.
kill_timestamp=None,
network_bytes_sent=0,
next_outbound_id=1,
next_processed_request=1,
next_states=set(),
output=output_collection,
# Output collection is nullified when flow is terminated, so we're
# keeping the urn separately for further reference.
output_urn=(output_collection is not None)
and output_collection.urn,
outstanding_requests=0,
remaining_cpu_quota=args.cpu_limit,
state=rdfvalue.Flow.State.RUNNING,
# Have we sent a notification to the user.
user_notified=False,
)
# Store the context in the flow_obj for next time.
self.flow_obj.state.Register("context", context)
return context
