def main():
#global qtimes,qerr
usage="<program> times_to_extend_url_list processes iterations"
if len(sys.argv) < 4:
print("Not enough arguments. "+usage)
return
times_to_extend_url_list = int(sys.argv[1])
processes=int(sys.argv[2])
iterations=int(sys.argv[3])
urls = nextend(settings.URLS, times_to_extend_url_list)
random.shuffle(urls)
n = len(urls)
timeout_secs=max(n/10, 60)
logger = dry.logger.setup_log_size_rotating("log/urlclient_gevent_pool.log",
logname="urlclientgeventpool")
logger.info("START|times_to_extend_url_list:{i}|processes:{p}|timeout:{t}|urls:{n}|type:gevent".format(i=times_to_extend_url_list,
p=processes,
t=timeout_secs,
n=n))
pool = Pool(max(processes, 4))
qurltime = queue.Queue()
elapsed_time = []
for i in range(iterations):
start = time.time()
try:
jobs = [pool.spawn(load_url, url, logger, logstatus=True) for url in urls]
# jobs = [pool.spawn(load_url, url, logger, timeout=60, logstatus=True) for url in urls]
pool.join(timeout=timeout_secs)
except Exception as e:
logger.error(e)
elapsed_time.append(time.time() - start)
elapsed_time_str = ""
for t in elapsed_time:
elapsed_time_str += str(t)+","
elapsed_time_str.strip,(",")
print("g,{u},{np},{et}".format(u=len(urls), np=processes, et=elapsed_time_str))
def main():
start = time.time()
logger = dry.logger.setup_log_size_rotating("log/bigfile_futures_threadpool.log",
logname='bigfilefuturesthreads')
logger.info("START")
elapsed_time = []
sfile = settings.BIG_FILE
fsize = os.path.getsize(sfile)
with open(sfile, "r") as fh:
#A list of tuples (chunk_start, chunk_size)
chunks = size_chunks(fh, fsize, num_chunks=settings.BIGFILE_FUTURES_CHUNKS)
pattern = re.compile(settings.TARGET_USERNAME)
file_handles = []
for j in range(len(chunks)):
file_handles.append(open(sfile, "r"))
with futures.ThreadPoolExecutor(max_workers=settings.BIGFILE_FUTURES_CHUNKS) as executor:
future_to_chunk = dict( (executor.submit(find_noq, file_handles[i], chunks[i], pattern), "") \
for i in range(len(chunks)) )
recsmatch = 0
try:
for future in futures.as_completed(future_to_chunk, timeout=60):
recsmatch += future.result()
except Exception as e:
#traceback.print_exc(file=sys.stdout)
logger.error("recsmatch={m} e={e}".format(m=recsmatch, e=e))
return
elapsed_time.append(time.time() - start)
elapsed_time_str = ""
for t in elapsed_time:
elapsed_time_str += str(t)+","
elapsed_time_str.strip,(",")
print("{r}".format(r=recsmatch))
logger.info("STOP|elapsedtime:{et}|recsmatch:{r}".format(et=elapsed_time, r=recsmatch))
def main():
usage="urlclient_futures t|p times_to_extend_url_list processes iterations"
if len(sys.argv) < 5:
print("Not enough arguments. "+usage)
return
if sys.argv[1] .lower() == "t":
process_type="threads"
elif sys.argv[1].lower() == "p":
process_type = "mp"
else:
print("Must specify t or p for thread or process as first argument.")
print(usage)
return
times_to_extend_url_list = int(sys.argv[2])
processes=int(sys.argv[3])
iterations = int(sys.argv[4])
urls = nextend(settings.URLS, times_to_extend_url_list)
n = len(urls)
timeout_secs=max(n*3, 60)
logger = dry.logger.setup_log_size_rotating("log/urlclient_futures.log",
logname='urlclientfutures')
logger.info("times_to_extend_url_list={i} processes={p} timeout={t} n={n} type={y}".format(i=times_to_extend_url_list,
p=processes,
t=timeout_secs,
n=n,
y=process_type))
elapsed_time = []
for i in range(iterations):
j = 0
start = time.time()
if process_type == "threads":
with futures.ThreadPoolExecutor(max_workers=processes) as executor:
future_to_url = dict( (executor.submit(load_url, url, logger, timeout_secs=30,logstatus=True), url) for url in urls )
else:
#urls=urls[:9]
try:
with futures.ProcessPoolExecutor(max_workers=processes) as executor:
future_to_url = dict( (executor.submit(load_url_mp, url, timeout_secs=10), url) for url in urls )
except Exception as e:
print(e)
#logger.error(e)
try:
for future in futures.as_completed(future_to_url, timeout=timeout_secs):
j += 1
url = future_to_url[future]
except Exception as e:
#traceback.print_exc(file=sys.stdout)
logger.error("j={j} url={u} e={e}".format(j=j, e=e, u=url))
elapsed_time.append(time.time() - start)
elapsed_time_str = ""
for t in elapsed_time:
elapsed_time_str += str(t)+","
elapsed_time_str.strip,(",")
print("{pt},{u},{np},{et}".format(pt=process_type, u=len(urls), np=processes,
et=elapsed_time_str))