3 ###############################################################################
6 # Copyright (C) 2014 INRIA
8 # This program is free software: you can redistribute it and/or modify
9 # it under the terms of the GNU General Public License version 2 as
10 # published by the Free Software Foundation;
12 # This program is distributed in the hope that it will be useful,
13 # but WITHOUT ANY WARRANTY; without even the implied warranty of
14 # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
15 # GNU General Public License for more details.
17 # You should have received a copy of the GNU General Public License
18 # along with this program. If not, see <http://www.gnu.org/licenses/>.
21 # Author: Alina Quereilhac <alina.quereilhac@inria.fr>
23 ###############################################################################
26 # This library contains functions to parse (CCNx) ccnd logs.
28 # Results from experiments must be stored in a directory
29 # named with the experiment run id.
30 # ccnd logs are stored in .log files in a subdirectory per node.
31 # The following diagram exemplifies the experiment result directory
32 # structure (nidi is the unique identifier assigned to node i):
43 from __future__ import print_function
52 from nepi.util.timefuncs import compute_delay_ms
53 from nepi.util.statfuncs import compute_mean
54 import nepi.data.processing.ping.parser as ping_parser
56 def is_control(content_name):
57 return content_name.startswith("ccnx:/%C1") or \
58 content_name.startswith("ccnx:/ccnx") or \
59 content_name.startswith("ccnx:/...")
62 def parse_file(filename):
63 """ Parses message information from ccnd log files
65 filename: path to ccndlog file
72 f = open(filename, "r")
77 cols = line.strip().split(sep)
80 # MESSAGE interest_from
81 # 1374181938.808523 ccnd[9245]: debug.4352 interest_from 6 ccnx:/test/bunny.ts (23 bytes,sim=0CDCC1D7)
84 # 1374181938.812750 ccnd[9245]: debug.3502 interest_to 5 ccnx:/test/bunny.ts (39 bytes,i=2844,sim=0CDCC1D7)
86 # MESSAGE CONTENT FROM
87 # 1374181938.868682 ccnd[9245]: debug.4643 content_from 5 ccnx:/test/bunny.ts/%FD%05%1E%85%8FVw/%00/%9E%3D%01%D9%3Cn%95%2BvZ%8
90 # 1374181938.868772 ccnd[9245]: debug.1619 content_to 6 ccnx:/test/bunny.ts/%FD%05%1E%85%8FVw/%00/%9E%3D%01%D9%3Cn%95%2BvZ%8
92 # 1375596708.222304 ccnd[9758]: debug.3692 interest_expiry ccnx:/test/bunny.ts/%FD%05%1E%86%B1GS/%00%0A%F7 (44 bytes,c=0:1,i=2819,sim=49FA8048)
94 # External face creation
95 # 1374181452.965961 ccnd[9245]: accepted datagram client id=5 (flags=0x40012) 204.85.191.10 port 9695
97 if line.find("accepted datagram client") > -1:
98 face_id = (cols[5]).replace("id=",'')
101 faces[face_id] = (ip, port)
104 # 1374181452.985296 ccnd[9245]: releasing face id 4 (slot 4)
105 if line.find("releasing face id") > -1:
115 message_type = cols[3]
117 if message_type not in ["interest_from", "interest_to", "content_from",
118 "content_to", "interest_dupnonce", "interest_expiry"]:
122 content_name = cols[5]
124 # Interest Nonce ? -> 412A74-0844-0008-50AA-F6EAD4
126 if message_type in ["interest_from", "interest_to", "interest_dupnonce"]:
128 if len(last.split("-")) == 5:
132 size = int((cols[6]).replace('(',''))
134 print("interest_expiry without face id!", line)
137 # If no external IP address was identified for this face
138 # asume it is a local face
142 peer, port = faces[face_id]
144 data.append((content_name, timestamp, message_type, peer, face_id,
151 def dump_content_history(content_history):
152 f = tempfile.NamedTemporaryFile(delete=False)
153 pickle.dump(content_history, f)
157 def load_content_history(fname):
159 content_history = pickle.load(f)
163 return content_history
165 def annotate_cn_node(graph, nid, ips2nid, data, content_history):
166 for (content_name, timestamp, message_type, peer, face_id,
167 size, nonce, line) in data:
169 # Ignore control messages for the time being
170 if is_control(content_name):
173 if message_type == "interest_from" and \
175 graph.node[nid]["ccn_consumer"] = True
176 elif message_type == "content_from" and \
178 graph.node[nid]["ccn_producer"] = True
180 # Ignore local messages for the time being.
181 # They could later be used to calculate the processing times
183 if peer == "localhost":
187 if message_type in ["content_from", "content_to"]:
188 content_name = "/".join(content_name.split("/")[:-1])
190 if content_name not in content_history:
191 content_history[content_name] = list()
193 peernid = ips2nid[peer]
194 graph.add_edge(nid, peernid)
196 content_history[content_name].append((timestamp, message_type, nid,
197 peernid, nonce, size, line))
199 def annotate_cn_graph(logs_dir, graph, parse_ping_logs = False):
200 """ Adds CCN content history for each node in the topology graph.
204 # Make a copy of the graph to ensure integrity
209 for nid in graph.nodes():
210 ips = graph.node[nid]["ips"]
216 # Now walk through the ccnd logs...
217 for dirpath, dnames, fnames in os.walk(logs_dir):
218 # continue if we are not at the leaf level (if there are subdirectories)
222 # Each dirpath correspond to a different node
223 nid = os.path.basename(dirpath)
225 # Cast to numeric nid if necessary
226 if int(nid) in graph.nodes():
229 content_history = dict()
232 if fname.endswith(".log"):
234 filename = os.path.join(dirpath, fname)
235 data = parse_file(filename)
236 annotate_cn_node(graph, nid, ips2nid, data, content_history)
238 # Avoid storing everything in memory, instead dump to a file
239 # and reference the file
240 fname = dump_content_history(content_history)
241 graph.node[nid]["history"] = fname
244 msg = "No CCND output files were found to parse at %s " % logs_dir
245 raise RuntimeError, msg
248 ping_parser.annotate_cn_graph(logs_dir, graph)
252 def ccn_producers(graph):
253 """ Returns the nodes that are content providers """
254 return [nid for nid in graph.nodes() \
255 if graph.node[nid].get("ccn_producer")]
257 def ccn_consumers(graph):
258 """ Returns the nodes that are content consumers """
259 return [nid for nid in graph.nodes() \
260 if graph.node[nid].get("ccn_consumer")]
262 def process_content_history(graph):
263 """ Compute CCN message counts and aggregates content historical
264 information in the content_names dictionary
268 ## Assume single source
269 source = ccn_consumers(graph)[0]
271 interest_expiry_count = 0
272 interest_dupnonce_count = 0
275 content_names = dict()
277 # Collect information about exchanged messages by content name and
279 for nid in graph.nodes():
280 # Load the data collected from the node's ccnd log
281 fname = graph.node[nid]["history"]
282 history = load_content_history(fname)
284 for content_name in history.keys():
285 hist = history[content_name]
287 for (timestamp, message_type, nid1, nid2, nonce, size, line) in hist:
288 if message_type in ["content_from", "content_to"]:
289 # The first Interest sent will not have a version or chunk number.
290 # The first Content sent back in reply, will end in /=00 or /%00.
291 # Make sure to map the first Content to the first Interest.
292 if content_name.endswith("/=00"):
293 content_name = "/".join(content_name.split("/")[0:-2])
295 # Add content name to dictionary
296 if content_name not in content_names:
297 content_names[content_name] = dict()
298 content_names[content_name]["interest"] = dict()
299 content_names[content_name]["content"] = list()
301 # Classify interests by replica
302 if message_type in ["interest_from"] and \
303 nonce not in content_names[content_name]["interest"]:
304 content_names[content_name]["interest"][nonce] = list()
306 # Add consumer history
308 if message_type in ["interest_to", "content_from"]:
309 # content name history as seen by the source
310 if "consumer_history" not in content_names[content_name]:
311 content_names[content_name]["consumer_history"] = list()
313 content_names[content_name]["consumer_history"].append(
314 (timestamp, message_type))
316 # Add messages per content name and cumulate totals by message type
317 if message_type == "interest_dupnonce":
318 interest_dupnonce_count += 1
319 elif message_type == "interest_expiry":
320 interest_expiry_count += 1
321 elif message_type == "interest_from":
323 # Append to interest history of the content name
324 content_names[content_name]["interest"][nonce].append(
325 (timestamp, nid2, nid1))
326 elif message_type == "content_from":
328 # Append to content history of the content name
329 content_names[content_name]["content"].append((timestamp, nid2, nid1))
335 # Compute the time elapsed between the time an interest is sent
336 # in the consumer node and when the content is received back
337 for content_name in content_names.keys():
338 # order content and interest messages by timestamp
339 content_names[content_name]["content"] = sorted(
340 content_names[content_name]["content"])
342 for nonce, timestamps in content_names[content_name][
343 "interest"].iteritems():
344 content_names[content_name]["interest"][nonce] = sorted(
347 history = sorted(content_names[content_name]["consumer_history"])
348 content_names[content_name]["consumer_history"] = history
350 # compute the rtt time of the message
352 waiting_content = False
353 interest_timestamp = None
354 content_timestamp = None
356 for (timestamp, message_type) in history:
357 if not waiting_content and message_type == "interest_to":
358 waiting_content = True
359 interest_timestamp = timestamp
362 if waiting_content and message_type == "content_from":
363 content_timestamp = timestamp
366 # If we can't determine who sent the interest, discard it
368 if interest_timestamp and content_timestamp:
369 rtt = compute_delay_ms(content_timestamp, interest_timestamp)
371 content_names[content_name]["rtt"] = rtt
372 content_names[content_name]["lapse"] = (interest_timestamp, content_timestamp)
376 interest_expiry_count,
377 interest_dupnonce_count,
381 def process_content_history_logs(logs_dir, graph, parse_ping_logs = False):
382 """ Parse CCN logs and aggregate content history information in graph.
383 Returns annotated graph and message countn and content names history.
386 ## Process logs and analyse data
388 graph = annotate_cn_graph(logs_dir, graph,
389 parse_ping_logs = parse_ping_logs)
391 print("Skipping: Error parsing ccnd logs", logs_dir)
394 source = ccn_consumers(graph)[0]
395 target = ccn_producers(graph)[0]
397 # Process the data from the ccnd logs, but do not re compute
402 interest_expiry_count,
403 interest_dupnonce_count,
405 content_count) = process_content_history(graph)
407 print("Skipping: Error processing ccn data", logs_dir)
412 interest_expiry_count,
413 interest_dupnonce_count,