Develop Reference

processing very large text files in parallel using multiprocessing and threading

I have found several other questions that touch on this topic but none that are quite like my situation.
I have several very large text files (3+ gigabytes in size).
I would like to process them (say 2 documents) in parallel using multiprocessing. As part of my processing (within a single process) I need to make an API call and because of this would like to have each process have it's own threads to run asynchronously.
I have came up with a simplified example ( I have commented the code to try to explain what I think it should be doing):
import multiprocessing
from threading import Thread
import threading
from queue import Queue
import time
def process_huge_file(*, file_, batch_size=250, num_threads=4):
# create APICaller instance for each process that has it's own Queue
api_call = APICaller()
batch = []
# create threads that will run asynchronously to make API calls
# I expect these to immediately block since there is nothing in the Queue (which is was
# the api_call.run depends on to make a call
threads = []
for i in range(num_threads):
thread = Thread(target=api_call.run)
threads.append(thread)
thread.start()
for thread in threads:
thread.join()
####
# start processing the file line by line
for line in file_:
# if we are at our batch size, add the batch to the api_call to to let the threads do
# their api calling
if i % batch_size == 0:
api_call.queue.put(batch)
else:
# add fake line to batch
batch.append(fake_line)
class APICaller:
def __init__(self):
# thread safe queue to feed the threads which point at instances
of these APICaller objects
self.queue = Queue()
def run(self):
print("waiting for something to do")
self.queue.get()
print("processing item in queue")
time.sleep(0.1)
print("finished processing item in queue")
if __name__ == "__main__":
# fake docs
fake_line = "this is a fake line of some text"
# two fake docs with line length == 1000
fake_docs = [[fake_line] * 1000 for i in range(2)]
####
num_processes = 2
procs = []
for idx, doc in enumerate(fake_docs):
proc = multiprocessing.Process(target=process_huge_file, kwargs=dict(file_=doc))
proc.start()
procs.append(proc)
for proc in procs:
proc.join()
As the code is now, "waiting for something to do" prints 8 times (makes sense 4 threads per process) and then it stops or "deadlocks" which is not what I expect - I expect it to start sharing time with the threads as soon as I start putting items in the Queue but the code does not appear to make it this far. I ordinarily would step through to find a hang up but I still don't have a solid understanding of how to best debug using Threads (another topic for another day).
In the meantime, can someone help me figure out why my code is not doing what it should be doing?

I have made a few adjustments and additions and the code appears to do what it is supposed to now. The main adjustments are: adding a CloseableQueue class (from Brett Slatkins Effective Python Item 55), and ensuring that I call close and join on the queue so that the threads properly exit. Full code with these changes below:
import multiprocessing
from threading import Thread
import threading
from queue import Queue
import time
from concurrency_utils import CloseableQueue
def sync_process_huge_file(*, file_, batch_size=250):
batch = []
for idx, line in enumerate(file_):
# do processing on the text
if idx % batch_size == 0:
time.sleep(0.1)
batch = []
# api_call.queue.put(batch)
else:
computation = 0
for i in range(100000):
computation += i
batch.append(line)
def process_huge_file(*, file_, batch_size=250, num_threads=4):
api_call = APICaller()
batch = []
# api call threads
threads = []
for i in range(num_threads):
thread = Thread(target=api_call.run)
threads.append(thread)
thread.start()
for idx, line in enumerate(file_):
# do processing on the text
if idx % batch_size == 0:
api_call.queue.put(batch)
else:
computation = 0
for i in range(100000):
computation += i
batch.append(line)
for _ in threads:
api_call.queue.close()
api_call.queue.join()
for thread in threads:
thread.join()
class APICaller:
def __init__(self):
self.queue = CloseableQueue()
def run(self):
for item in self.queue:
print("waiting for something to do")
pass
print("processing item in queue")
time.sleep(0.1)
print("finished processing item in queue")
print("exiting run")
if __name__ == "__main__":
# fake docs
fake_line = "this is a fake line of some text"
# two fake docs with line length == 1000
fake_docs = [[fake_line] * 10000 for i in range(2)]
####
time_s = time.time()
num_processes = 2
procs = []
for idx, doc in enumerate(fake_docs):
proc = multiprocessing.Process(target=process_huge_file, kwargs=dict(file_=doc))
proc.start()
procs.append(proc)
for proc in procs:
proc.join()
time_e = time.time()
print(f"took {time_e-time_s} ")
class CloseableQueue(Queue):
SENTINEL = object()
def __init__(self, **kwargs):
super().__init__(**kwargs)
def close(self):
self.put(self.SENTINEL)
def __iter__(self):
while True:
item = self.get()
try:
if item is self.SENTINEL:
return # exit thread
yield item
finally:
self.task_done()
As expected this is a great speedup from running synchronously - 120 seconds vs 50 seconds.

How to log to single file with multiprocessing.Pool.apply_async

I can't get logging to a single file working with multprocess.Pool.apply_async.
I'm trying to adapt this example from the Logging Cookbook, but it only works for multiprocessing.Process. Passing the logging queue into apply_async doesn't seem to have effect.
I would like to use a Pool so that I can easily manage the number of simultaneous threads.
The following adapted example with multiprocessing.Process works ok for me, except I am not getting log messages from the main process, and I don't think it will work well when I have 100 large jobs.
import logging
import logging.handlers
import numpy as np
import time
import multiprocessing
import pandas as pd
log_file = 'PATH_TO_FILE/log_file.log'
def listener_configurer():
root = logging.getLogger()
h = logging.FileHandler(log_file)
f = logging.Formatter('%(asctime)s %(processName)-10s %(name)s %(levelname)-8s %(message)s')
h.setFormatter(f)
root.addHandler(h)
# This is the listener process top-level loop: wait for logging events
# (LogRecords)on the queue and handle them, quit when you get a None for a
# LogRecord.
def listener_process(queue, configurer):
configurer()
while True:
try:
record = queue.get()
if record is None: # We send this as a sentinel to tell the listener to quit.
break
logger = logging.getLogger(record.name)
logger.handle(record) # No level or filter logic applied - just do it!
except Exception:
import sys, traceback
print('Whoops! Problem:', file=sys.stderr)
traceback.print_exc(file=sys.stderr)
def worker_configurer(queue):
h = logging.handlers.QueueHandler(queue) # Just the one handler needed
root = logging.getLogger()
root.addHandler(h)
# send all messages, for demo; no other level or filter logic applied.
root.setLevel(logging.DEBUG)
# This is the worker process top-level loop, which just logs ten events with
# random intervening delays before terminating.
# The print messages are just so you know it's doing something!
def worker_function(sleep_time, name, queue, configurer):
configurer(queue)
start_message = 'Worker {} started and will now sleep for {}s'.format(name, sleep_time)
logging.info(start_message)
time.sleep(sleep_time)
success_message = 'Worker {} has finished sleeping for {}s'.format(name, sleep_time)
logging.info(success_message)
def main_with_process():
start_time = time.time()
single_thread_time = 0.
queue = multiprocessing.Queue(-1)
listener = multiprocessing.Process(target=listener_process,
args=(queue, listener_configurer))
listener.start()
workers = []
for i in range(10):
name = str(i)
sleep_time = np.random.randint(10) / 2
single_thread_time += sleep_time
worker = multiprocessing.Process(target=worker_function,
args=(sleep_time, name, queue, worker_configurer))
workers.append(worker)
worker.start()
for w in workers:
w.join()
queue.put_nowait(None)
listener.join()
end_time = time.time()
final_message = "Script execution time was {}s, but single-thread time was {}s".format(
(end_time - start_time),
single_thread_time
)
print(final_message)
if __name__ == "__main__":
main_with_process()
But I can't get the following adaptation to work:
def main_with_pool():
start_time = time.time()
queue = multiprocessing.Queue(-1)
listener = multiprocessing.Process(target=listener_process,
args=(queue, listener_configurer))
listener.start()
pool = multiprocessing.Pool(processes=3)
job_list = [np.random.randint(10) / 2 for i in range(10)]
single_thread_time = np.sum(job_list)
for i, sleep_time in enumerate(job_list):
name = str(i)
pool.apply_async(worker_function,
args=(sleep_time, name, queue, worker_configurer))
queue.put_nowait(None)
listener.join()
end_time = time.time()
print("Script execution time was {}s, but single-thread time was {}s".format(
(end_time - start_time),
single_thread_time
))
if __name__ == "__main__":
main_with_pool()
I've tried many slight variations, using multiprocessing.Manager, multiprocessing.Queue, multiprocessing.get_logger, apply_async.get(), but haven't gotten any to work.
I would think there would be an off-the-shelf solution for this. Should I try Celery instead?
thanks

There are actually two separate problems here, which are intertwined:
You cannot pass a multiprocessing.Queue() object as an argument to a Pool-based function (you can pass it to the worker you start directly, but not any "further in" as it were).
You must wait for all the asynchronous workers to complete before you send the None through to your listener process.
To fix the first one, replace:
queue = multiprocessing.Queue(-1)
with:
queue = multiprocessing.Manager().Queue(-1)
as a manager-managed Queue() instance can be passed through.
To fix the second, either collect each result from each asynchronous call, or close the pool and wait for it, e.g.:
pool.close()
pool.join()
queue.put_nowait(None)
or the more complex:
getters = []
for i, sleep_time in enumerate(job_list):
name = str(i)
getters.append(
pool.apply_async(worker_function,
args=(sleep_time, name, queue, worker_configurer))
)
while len(getters):
getters.pop().get()
# optionally, close and join pool here (generally a good idea anyway)
queue.put_nowait(None)
(You should also consider replacing your put_nowait with a waiting version of put and not using unlimited length queues.)

Consider using two queues. The first queue is where you put the data for the workers. Each worker after job completion pushes the results to the second queue. Now consume this second queue to write the log to the file.

[ADDENDUM] Regarding maxtasksperchild=1
you don't really need it. The reason for repeated messages were due to:
you were repeatedly adding queuehandlers to the root logger of a child process. The following code checks if any handlers exist before adding another:
def worker_configurer(queue):
root = logging.getLogger()
# print(f'{root.handlers=}')
if len(root.handlers) == 0:
h = logging.handlers.QueueHandler(queue)
root.addHandler(h)
root.setLevel(logging.DEBUG)

Real-time-ability python multiprocessing (Queue and Pipe)

I am a little bit confused testing the multiprocessing module.
Let's simulate a digital timer. The code would look like:
start=datetime.now()
while True:
now=datetime.now()
delta=now-start
s = delta.seconds + delta.microseconds/1E6
print s
time.sleep(1)
Which returns correctly:
8e-06
1.001072
2.00221
3.003353
4.004416
...
Now I want to read the clock from my virtual external digital clock device using a pipe:
def ask_timer(conn):
start=datetime.now()
while True:
now=datetime.now()
delta=now-start
s = delta.seconds + delta.microseconds/1E6
conn.send(s)
parent_conn, child_conn = Pipe()
p = Process(target=ask_timer, args=(child_conn,))
p.start()
while True:
print parent_conn.recv()
time.sleep(1)
It returns:
2.9e-05
6.7e-05
7.7e-05
8.3e-05
8.9e-05
9.4e-05
0.0001
...
Here the timer doesn't seem to run permanently in the background..The implementation of "Queue" looks like:
def ask_timer(q):
while True:
now=datetime.now()
delta=now-start
s = delta.seconds + delta.microseconds/1E6
q.put(s)
#conn.close()
q = Queue()
p = Process(target=ask_timer, args=(q,))
p.start()
while True:
print q.get()
time.sleep(1)
which does the same like pipe. Is this just my misconception of multiprocessing of python? How could I ask a value realtime from a running parallel-thread?

Everything is working correctly. The child process is executing ask_timer() function completely independently from you main process. You don't have any time.sleep() in this function, so it just prints or puts in the queue deltas in the infinite loop with interval of like 10ms.
Once a second your main process asks child process for data and get's it. Data is one of those small intervals.
The problem there is that you're putting much more data into pipe/queue, than taking from it. So you're getting old data, when you ask. To test that you can print queue size in the loop (won't work on OS X):
def ask_timer(q):
start = datetime.now()
while True:
now = datetime.now()
delta = now - start
s = delta.seconds + delta.microseconds / 1E6
q.put(s)
q = Queue()
p = Process(target=ask_timer, args=(q,))
p.start()
while True:
print q.get()
print q.qsize()
time.sleep(1)
The queue size will grow really fast.
Apparently you can use shared memory to read current value from the child process.
from multiprocessing import Process, Value
from datetime import datetime
import time
from ctypes import c_double
def ask_timer(v):
start = datetime.now()
while True:
now = datetime.now()
delta = now - start
s = delta.seconds + delta.microseconds / 1E6
v.value = s
val = Value(c_double, 0.0)
p = Process(target=ask_timer, args=(val,))
p.start()
while True:
print(val.value)
time.sleep(1)

kill threads later a time in python

I have a python code with threads, and i need that if in for example 1 hour the threads are not finished, finish all threads and finish the script, and if the hour are not complete wait that all my threads finish.
I try with a daemon thread, and with a sleep of the hour, and if the hour is complete use a: sys.exit() but it not works to me, because always wait to my sleep threadh, then my script wait until the thread finished and the sys.exit() does not work.
import socket, threading, time, sys
from sys import argv
import os
acc_time=0
transactions_ps=5
ins = open(sys.argv[1],'r')
msisdn_list = []
for line in ins:
msisdn_list.append (line.strip('\n'))
# print line
ins.close()
def worker(msisdn_list):
semaphore.acquire()
global transactions_ps
print " ***** ", threading.currentThread().getName(), "Lanzado"
count=1
acc_time=0
print "len: ",len(msisdn_list)
for i in msisdn_list:
try:
init=time.time()
time.sleep(2)
print "sleeping...",i
time.sleep(4)
final=time.time()
acc_time = acc_time+final-init
print acc_time
except IOError:
print "Connection failed",sys.exc_info()[0]
print "Deteniendo ",threading.currentThread().getName()
semaphore.release()
def kill_process(secs_to_die):
time.sleep(secs_to_die)
sys.exit()
seconds_to_die=3600
thread_kill = threading.Thread(target = kill_process, args=(seconds_to_die,))
thread_kill.start()
max_con=5
semaphore = threading.BoundedSemaphore(max_con)
for i in range(0,28,transactions_ps):
w = threading.Thread(target=worker, args=(msisdn_list[i:i+transactions_ps-1],))
w.setDaemon(True)
w.start()
How can to do it

A minimal change to your code that would fix the issue is threading.Barrier:
barrier = Barrier(number_of_threads, timeout=3600)
# create (number_of_threads - 1) threads, pass them barrier
# each thread calls barrier.wait() on exit
barrier.wait() # after number_of_threads .wait() calls or on timeout it returns
A simpler alternative is to use multiprocessing.dummy.Pool that creates daemon threads:
from multiprocessing.dummy import Pool # use threads
start = timer()
endtime = start + 3600
for result in pool.imap_unordered(work, args):
if timer() > endtime:
exit("timeout")
The code doesn't timeout until a work item is done i.e., it expects that processing a single item from the list doesn't take long.
Complete example:
#!/usr/bin/env python3
import logging
import multiprocessing as mp
from multiprocessing.dummy import Pool
from time import monotonic as timer, sleep
info = mp.get_logger().info
def work(i):
info("start %d", i)
sleep(1)
info("end %d", i)
seconds_to_die = 3600
max_con = 5
mp.log_to_stderr().setLevel(logging.INFO) # enable logging
pool = Pool(max_con) # no more than max_con at a time
start = timer()
endtime = start + seconds_to_die
for _ in pool.imap_unordered(work, range(10000)):
if timer() > endtime:
exit("timeout")

You may refer to this implementation of KThread:
http://python.todaysummary.com/q_python_45717.html

Threads not stop in python

The purpose of my program is to download files with threads. I define the unit, and using len/unit threads, the len is the length of the file which is going to be downloaded.
Using my program, the file can be downloaded, but the threads are not stopping. I can't find the reason why.
This is my code...
#! /usr/bin/python
import urllib2
import threading
import os
from time import ctime
class MyThread(threading.Thread):
def __init__(self,func,args,name=''):
threading.Thread.__init__(self);
self.func = func;
self.args = args;
self.name = name;
def run(self):
apply(self.func,self.args);
url = 'http://ubuntuone.com/1SHQeCAQWgIjUP2945hkZF';
request = urllib2.Request(url);
response = urllib2.urlopen(request);
meta = response.info();
response.close();
unit = 1000000;
flen = int(meta.getheaders('Content-Length')[0]);
print flen;
if flen%unit == 0:
bs = flen/unit;
else :
bs = flen/unit+1;
blocks = range(bs);
cnt = {};
for i in blocks:
cnt[i]=i;
def getStr(i):
try:
print 'Thread %d start.'%(i,);
fout = open('a.zip','wb');
fout.seek(i*unit,0);
if (i+1)*unit > flen:
request.add_header('Range','bytes=%d-%d'%(i*unit,flen-1));
else :
request.add_header('Range','bytes=%d-%d'%(i*unit,(i+1)*unit-1));
#opener = urllib2.build_opener();
#buf = opener.open(request).read();
resp = urllib2.urlopen(request);
buf = resp.read();
fout.write(buf);
except BaseException:
print 'Error';
finally :
#opener.close();
fout.flush();
fout.close();
del cnt[i];
# filelen = os.path.getsize('a.zip');
print 'Thread %d ended.'%(i),
print cnt;
# print 'progress : %4.2f'%(filelen*100.0/flen,),'%';
def main():
print 'download at:',ctime();
threads = [];
for i in blocks:
t = MyThread(getStr,(blocks[i],),getStr.__name__);
threads.append(t);
for i in blocks:
threads[i].start();
for i in blocks:
# print 'this is the %d thread;'%(i,);
threads[i].join();
#print 'size:',os.path.getsize('a.zip');
print 'download done at:',ctime();
if __name__=='__main__':
main();
Could someone please help me understand why the threads aren't stopping.

I can't really address your code example because it is quite messy and hard to follow, but a potential reason you are seeing the threads not end is that a request will stall out and never finish. urllib2 allows you to specify timeouts for how long you will allow the request to take.
What I would recommend for your own code is that you split your work up into a queue, start a fixed number of thread (instead of a variable number), and let the worker threads pick up work until it is done. Make the http requests have a timeout. If the timeout expires, try again or put the work back into the queue.
Here is a generic example of how to use a queue, a fixed number of workers and a sync primitive between them:
import threading
import time
from Queue import Queue
def worker(queue, results, lock):
local_results = []
while True:
val = queue.get()
if val is None:
break
# pretend to do work
time.sleep(.1)
local_results.append(val)
with lock:
results.extend(local_results)
print threading.current_thread().name, "Done!"
num_workers = 4
threads = []
queue = Queue()
lock = threading.Lock()
results = []
for i in xrange(100):
queue.put(i)
for _ in xrange(num_workers):
# Use None as a sentinel to signal the threads to end
queue.put(None)
t = threading.Thread(target=worker, args=(queue,results,lock))
t.start()
threads.append(t)
for t in threads:
t.join()
print sorted(results)
print "All done"