/
task.py
344 lines (317 loc) · 12.6 KB
/
task.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
import socket
import threading
import struct
import cPickle as pickle
import logging
import beans
import sys
import marshal
import time
#----
import params
from util import TcpHelper as helper
from dfsclient import DfsClient
class JobInfo:
def __init__(self, id, maxM, mapR, ip, port):
self.id = id
self.max_map = maxM
self.max_reduce = mapR
self.map_running=[]
self.map_finished = []
self.reduce_running = []
self.reduce_finished=[]
self.ip = ip
self.file_server_port = port
def setMapFinished(self, map_id):
self.map_running.remove(map_id)
self.map_finished.append(map_id)
def addMapRunning(self, map_id):
self.map_running.append(map_id)
def setReduceFinished(self, map_id):
self.reduce_running.remove(map_id)
self.reduce_finished.append(map_id)
def addReduceRunning(self, map_id):
self.reduce_running.append(map_id)
def canAddMap(self):
if len(self.map_running) < self.max_map:
return True
return False
def canAddReduce(self):
if len(self.reduce_running) < self.max_reduce:
return True
return False
class Scheduler:
def __init__(self, source):
self.source = source
self.length = len(source)
self.iter = iter(source)
self.running = {}
self.finished = {}
self.id_pair = {}#work(map, reduce)id, job id pair
self.THRESHOLD = 2.0
num = self.length % 3
self.LEAST_FINISH = num if num>2 else self.length - 1
def next_task(self):
try:
key = self.iter.next()
return key
except:
#all were assigned, but only finish a little, wait
if len(self.finished) < self.LEAST_FINISH:
return -1
#re assign work that is too slow
for (k,t) in self.running.items():
if (time.time() - t) > self.THRESHOLD * self._getAverageTime():
logging.info('reassign work %d' % k)
return k
return -1
def isDown(self):
if len(self.finished) != self.length:
return False
return True
def setRunning(self, id):
self.running[id] = time.time()
def setFinished(self, id):
self.finished[id] = time.time() - self.running[id]
del self.running[id]
def _getAverageTime(self):
count = len(self.finished)
sum = 0
for k in self.finished:
sum += self.finished[k]
return float(sum) / count
def addIdPair(self, kid, vid):
self.id_pair[kid] = vid
class Task:
START = 0x01
MAP = 1
REDUCE = 2
ALLDOWN = 3
DISK_MODE = 0x01
MEM_MODE = 0x02
def __init__(self):
self.datasource = None
#for task info
self.reducer_num = 2
self.mapper_num = None
self.task_info = None
self.mapfn = None
self.reducefn = None
self.combinefn = None
self.dfsfile = None
self.DATA_TYPE = Task.MEM_MODE
#for scheduel
self.jobs = {}
self.map_scheduler = None
self.map_result = {}
self.job_id = 0
self.state = Task.START
self.reduce_scheduler = Scheduler(range(self.reducer_num))
#for secure
self.lock = threading.Lock()
def server_forever(self):
self.server = socket.socket(socket.AF_INET, socket.SOCK_STREAM)
self.server.bind(('', params.TASK_SERVER_PORT))
self.server.listen(100)
while True:
sock, addr = self.server.accept()
t=threading.Thread(target=self.handle, args=(sock,addr))
t.setDaemon(True)
t.start()
def handle(self, sock, addr):
data_len, = struct.unpack('!i', helper.receive_len(sock, 4))
all = helper.receive_len(sock, data_len)
pos = all.find(':')
cmd = str(all[: pos])
msg = all[pos+1:]
if cmd.startswith('job'):
self.handle_job(sock, addr, cmd, msg)
elif cmd.startswith('map'):
self.handle_map(sock, addr, cmd, msg)
elif cmd.startswith('reduce'):
self.handle_reduce(sock, addr, cmd, msg)
sock.close()
def handle_job(self, sock, addr, cmd, msg):
integer, = struct.unpack('!i', msg[0:4])
if cmd == 'job_start':
max_map = integer
max_reduce, port = struct.unpack('!2i', msg[4:12])
job = JobInfo(self.job_id, max_map, max_reduce, addr[0], port)
self.jobs[self.job_id] = job
logging.info('job client %d started' % self.job_id)
idB = struct.pack('!i', self.job_id)
self.send_command(sock, 'job_start', idB + self.task_info)
self.job_id += 1
elif cmd == 'job_heart':
jobId = integer
#logging.debug('heart beat form %d' % jobId)
#map phase
if self.state == Task.MAP:
if not self.map_scheduler.isDown():
if self.jobs[jobId].canAddMap():
id = self.map_scheduler.next_task()
if id >= 0:
logging.info('assigned map work %d' % id)
idB = struct.pack('!i', id)
with self.lock:
self.map_scheduler.addIdPair(id, jobId)
self.jobs[jobId].addMapRunning(id)
self.send_command(sock, 'job_map_assign', idB)
self.send_command(sock, 'job_idle')
else:
self.send_command(sock, 'job_idle')
self.state = Task.REDUCE
logging.info('start reducing...')
#reduce phase
elif self.state == Task.REDUCE:
if not self.reduce_scheduler.isDown():
if self.jobs[jobId].canAddReduce():
id = self.reduce_scheduler.next_task()
if id >= 0:
logging.info('assigned reduce work %d' % id)
idB = struct.pack('!i', id)
with self.lock:
self.reduce_scheduler.addIdPair(id, jobId)
self.jobs[jobId].addReduceRunning(id)
self.send_command(sock, 'job_reduce_assign', idB)
self.send_command(sock, 'job_idle')
else:
self.send_command(sock, 'job_idle')
self.state = Task.ALLDOWN
logging.info('reducing down, task finished...')
elif self.state == Task.ALLDOWN:
self.send_command(sock, 'job_exit')
def handle_map(self, sock, addr, cmd, msg):
id, = struct.unpack('!i', msg[0:4])
if cmd == 'map_start':
logging.info('map %d start connect...' % id)
with self.lock:
self.map_scheduler.setRunning(id)
source = pickle.dumps(self.map_scheduler.source[id], True)
self.send_command(sock, 'map_start', source)
elif cmd=='map_down':
with self.lock:
self.map_scheduler.setFinished(id)
t = self.map_scheduler.finished[id]
jobid = self.map_scheduler.id_pair[id]
self.jobs[jobid].setMapFinished(id)
logging.info('map %d down %f, %d/%d' % (id, t, len(self.map_scheduler.finished), self.map_scheduler.length))
self.send_command(sock, 'map_exit')
elif cmd =='':
pass
def handle_reduce(self, sock, addr, cmd, msg):
id, = struct.unpack('!i', msg[0:4])
if cmd == 'reduce_start':
logging.info('reduce %d start connect...' % id)
with self.lock:
self.reduce_scheduler.setRunning(id)
file_servers = []
for (k,v) in self.jobs.items():
file_servers.append((v.ip, v.file_server_port, v.map_finished))
data = pickle.dumps(file_servers, True)
self.send_command(sock, 'reduce_start', data)
elif cmd=='reduce_down':
with self.lock:
self.reduce_scheduler.setFinished(id)
t = self.reduce_scheduler.finished[id]
jobid = self.reduce_scheduler.id_pair[id]
self.jobs[jobid].setReduceFinished(id)
logging.info('reduce %d down %f, %d/%d' % (id, t, len(self.reduce_scheduler.finished), self.reducer_num))
self.send_command(sock, 'reduce_exit')
elif cmd =='':
pass
'''
task info send to jobclient, so the jobclient parse in this format
data format:
1.1.map_num: int 4byte
1.2 reduce_num: int 4byte
2.mapfn_len : int 4byte
3.mapfn: N byte
4.reducefn_len : int 4byte
5.reducefn: N byte
6.is combiner exists: 0x01 exist
7.combinefn_len : int 4byte
8.combinefn: N byte
9.job_mode: 1 byte, (0x01, mem) (0x02, disk)
10.dfsfile name len: int 4byte (mode = 0x02)
11. dfsfile name: N byte
'''
def _init_task_info(self):
data = []
map_reduce_num = struct.pack('!2i', self.mapper_num, self.reducer_num)
data.append(map_reduce_num)
#map
mapfn = marshal.dumps(self.mapfn.func_code)
mapfn_len = len(mapfn)
mapfn_len_data = struct.pack('!i', mapfn_len)
data.extend([mapfn_len_data, mapfn])
#reduce
reducefn = marshal.dumps(self.reducefn.func_code)
reducefn_len = len(reducefn)
reducefn_len_data = struct.pack('!i', reducefn_len)
data.extend([reducefn_len_data, reducefn])
#combine
if self.combinefn is not None:
isExist = struct.pack('!b', 0x01)
combinefn = marshal.dumps(self.combinefn.func_code)
combinefn_len = len(combinefn)
combinefn_len_data = struct.pack('!i', combinefn_len)
data.extend([isExist, combinefn_len_data, combinefn])
else:
isExist = struct.pack('!b', 0x02)
data.append(isExist)
#job mode, dfsfile
if self.DATA_TYPE == Task.DISK_MODE:
modeB = struct.pack('!b', Task.DISK_MODE)
name_len = len(self.dfsfile)
namelenB = struct.pack('!i', name_len)
data.extend([modeB, namelenB, self.dfsfile])
else:
modeB = struct.pack('!b', Task.MEM_MODE)
data.append(modeB)
self.task_info = ''.join(data)
def send_command(self, sock, cmd, msg=' '):
data = cmd + ':' + msg
data_len = struct.pack('!i', len(data))
sock.sendall(data_len + data)
def _process_datasource(self):
source = {}
if self.dfsfile is not None:
self.DATA_TYPE = Task.DISK_MODE
self.datasource = DfsClient().open(self.dfsfile)
cur = 0#map id
for x in self.datasource.blocks:
source[cur] = x
cur += 1
else:
self.DATA_TYPE = Task.MEM_MODE
#segment the memory datasource into block,block size according to params
block = []
cur = 0#map id
start, end = 0, 0
for x in self.datasource:
block.append(x)
if sys.getsizeof(block)> params.TASK_MAP_MEM_BLOCK:
source[cur] = block
block = []
cur += 1
source[cur] = block
self.map_scheduler = Scheduler(source)
self.mapper_num = self.map_scheduler.length
def start(self):
self._process_datasource()
self._init_task_info()
logging.info('total map count:%d, total reduce count:%d' % (self.map_scheduler.length, self.reducer_num))
logging.info('start mapping...')
self.state = self.MAP
t = threading.Thread(target=self.show_info)
t.start()
self.server_forever()
def show_info(self):
while True:
logging.info('running map %d, running reduce %d', \
len(self.map_scheduler.running), len(self.reduce_scheduler.running))
logging.info('task progress: map %d/%d, reduce %d/%d', \
len(self.map_scheduler.finished), self.mapper_num, len(self.reduce_scheduler.finished), self.reducer_num)
print ''
time.sleep(30)