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
|
# -*- coding: utf-8 -*-
# Copyright (C) 2014 Yahoo! Inc. All Rights Reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License"); you may
# not use this file except in compliance with the License. You may obtain
# a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS, WITHOUT
# WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the
# License for the specific language governing permissions and limitations
# under the License.
import collections
import contextlib
import logging
import os
import random
import sys
import threading
import time
logging.basicConfig(level=logging.ERROR)
top_dir = os.path.abspath(os.path.join(os.path.dirname(__file__),
os.pardir,
os.pardir))
sys.path.insert(0, top_dir)
from six.moves import range as compat_range
from zake import fake_client
from taskflow import exceptions as excp
from taskflow.jobs import backends
# In this example we show how a jobboard can be used to post work for other
# entities to work on. This example creates a set of jobs using one producer
# thread (typically this would be split across many machines) and then having
# other worker threads with there own jobboards select work using a given
# filters [red/blue] and then perform that work (and consuming or abandoning
# the job after it has been completed or failed).
# Things to note:
# - No persistence layer is used (or logbook), just the job details are used
# to determine if a job should be selected by a worker or not.
# - This example runs in a single process (this is expected to be atypical
# but this example shows that it can be done if needed, for testing...)
# - The iterjobs(), claim(), consume()/abandon() worker workflow.
# - The post() producer workflow.
SHARED_CONF = {
'path': "/taskflow/jobs",
'board': 'zookeeper',
}
# How many workers and producers of work will be created (as threads).
PRODUCERS = 3
WORKERS = 5
# How many units of work each producer will create.
PRODUCER_UNITS = 10
# How many units of work are expected to be produced (used so workers can
# know when to stop running and shutdown, typically this would not be a
# a value but we have to limit this examples execution time to be less than
# infinity).
EXPECTED_UNITS = PRODUCER_UNITS * PRODUCERS
# Delay between producing/consuming more work.
WORKER_DELAY, PRODUCER_DELAY = (0.5, 0.5)
# To ensure threads don't trample other threads output.
STDOUT_LOCK = threading.Lock()
def dispatch_work(job):
# This is where the jobs contained work *would* be done
time.sleep(1.0)
def safe_print(name, message, prefix=""):
with STDOUT_LOCK:
if prefix:
print("%s %s: %s" % (prefix, name, message))
else:
print("%s: %s" % (name, message))
def worker(ident, client, consumed):
# Create a personal board (using the same client so that it works in
# the same process) and start looking for jobs on the board that we want
# to perform.
name = "W-%s" % (ident)
safe_print(name, "started")
claimed_jobs = 0
consumed_jobs = 0
abandoned_jobs = 0
with backends.backend(name, SHARED_CONF.copy(), client=client) as board:
while len(consumed) != EXPECTED_UNITS:
favorite_color = random.choice(['blue', 'red'])
for job in board.iterjobs(ensure_fresh=True, only_unclaimed=True):
# See if we should even bother with it...
if job.details.get('color') != favorite_color:
continue
safe_print(name, "'%s' [attempting claim]" % (job))
try:
board.claim(job, name)
claimed_jobs += 1
safe_print(name, "'%s' [claimed]" % (job))
except (excp.NotFound, excp.UnclaimableJob):
safe_print(name, "'%s' [claim unsuccessful]" % (job))
else:
try:
dispatch_work(job)
board.consume(job, name)
safe_print(name, "'%s' [consumed]" % (job))
consumed_jobs += 1
consumed.append(job)
except Exception:
board.abandon(job, name)
abandoned_jobs += 1
safe_print(name, "'%s' [abandoned]" % (job))
time.sleep(WORKER_DELAY)
safe_print(name,
"finished (claimed %s jobs, consumed %s jobs,"
" abandoned %s jobs)" % (claimed_jobs, consumed_jobs,
abandoned_jobs), prefix=">>>")
def producer(ident, client):
# Create a personal board (using the same client so that it works in
# the same process) and start posting jobs on the board that we want
# some entity to perform.
name = "P-%s" % (ident)
safe_print(name, "started")
with backends.backend(name, SHARED_CONF.copy(), client=client) as board:
for i in compat_range(0, PRODUCER_UNITS):
job_name = "%s-%s" % (name, i)
details = {
'color': random.choice(['red', 'blue']),
}
job = board.post(job_name, book=None, details=details)
safe_print(name, "'%s' [posted]" % (job))
time.sleep(PRODUCER_DELAY)
safe_print(name, "finished", prefix=">>>")
def main():
with contextlib.closing(fake_client.FakeClient()) as c:
created = []
for i in compat_range(0, PRODUCERS):
p = threading.Thread(target=producer, args=(i + 1, c))
p.daemon = True
created.append(p)
p.start()
consumed = collections.deque()
for i in compat_range(0, WORKERS):
w = threading.Thread(target=worker, args=(i + 1, c, consumed))
w.daemon = True
created.append(w)
w.start()
while created:
t = created.pop()
t.join()
# At the end there should be nothing leftover, let's verify that.
board = backends.fetch('verifier', SHARED_CONF.copy(), client=c)
board.connect()
with contextlib.closing(board):
if board.job_count != 0 or len(consumed) != EXPECTED_UNITS:
return 1
return 0
if __name__ == "__main__":
sys.exit(main())
|
