如何在 Python 中的 concurrent.futures.ProcessPoolExecutor 中传递“锁”？

Question

I am running a parallel test with Python 3.7 and Appium 1.15.1 on real Android smartphones.

I use concurrent.futures.ProcessPoolExecutor to run each test on each smartphone.

I am passing list of the uid of smartphones to my map function. By this way, my method 'run_smartphone()'(which start the test) get the uid of smartphone and identify on which smartphone it must run the test.

My script work fine without any issue. But I would like to add a "lock" because 'run_smartphone()' make some I/O on a sqlite3 database. So correct me if I am wrong, but it would be a good practice to "lock" the I/O operation on this sqlite3 database?

Here is my original code which works:

def run_smartphone(p_udid):
    #do the stuff

list_smartphones_connected = [41492968379078, 53519716736397]
with concurrent.futures.ProcessPoolExecutor() as executor:
    try:
        multiprocesses = executor.map(mymodules.run_smartphone, list_smartphones_connected)

    except ValueError:
        print(("Error multiprocesses"))

So I tried to add pass "lock" to my method 'run_smartphone()'. This is what I wrote:

m = multiprocessing.Manager()
lock = m.Lock()
list_arguments_smartphones = []

list_smartphones_connected = [41492968379078, 53519716736397]
for smartphone_connected in list_smartphones_connected:        
    list_arguments_smartphones.append([smartphone_connected, lock])

with concurrent.futures.ProcessPoolExecutor() as executor:
    try:
        multiprocesses = executor.map(mymodules.run_smartphone, list_arguments_smartphones)

    except ValueError:
        print(("Error multiprocesses"))

But it doesn't work and I don't get any exception raised. Pycharm stop the script:

Process finished with exit code 0

I have no idea what is stopping the script.

So I started to investigate by executing the script for 1 smartphone with this line:

 multiprocesses = executor.map(mymodules.run_smartphone, [41492968379078,lock])

it gives the same result => The script stop, no automation start and I don't see any exception raised (Process finished with exit code 0).

As I wanted to know where exactly was the issue, I run the script with 'trace'.

py -m trace --trace  myscript.py

But I don't understand anything, I don't see any error... You can see the output of this 'trace' command on a text file I uploaded on GitHub:

https://github.com/gauthierbuttez/public/blob/master/trace-log.txt

Does anyone have any idea how can I pass the "lock" to my concurrent.futures.ProcessPoolExecutor()? And is it a good idea to do that?

Thanks.

Answer 1

May this help you...

    m = multiprocessing.Manager()
    lock = m.Lock()

    def run_smartphone(p_udid, lock): 
        # further code

    list_smartphones_connected = [41492968379078, 53519716736397] 
    with concurrent.futures.ProcessPoolExecutor() as executor: 
        try: 
            multiprocesses = executor.map(run_smartphone, list_smartphones_connected, [lock]*len(list_smartphones_connected)) 
            for function_return_value in multiprocesses:
                print(function_return_value)

        except ValueError: 
            print(("Error multiprocesses"))

Answer 2

From the docs for map():

If a func call raises an exception, then that exception will be raised when its value is retrieved from the iterator.

In other words, you probably have to actually use the return value of run_smartphone:

m = multiprocessing.Manager()
lock = m.Lock()
list_arguments_smartphones = []

list_smartphones_connected = [41492968379078, 53519716736397]
for smartphone_connected in list_smartphones_connected:        
    list_arguments_smartphones.append([smartphone_connected, lock])

with concurrent.futures.ProcessPoolExecutor() as executor:
    try:
        multiprocesses = executor.map(mymodules.run_smartphone, list_arguments_smartphones)

        for function_return_value in multiprocesses:
            print(function_return_value)
            # or do something with the value, like insert into a db

    except ValueError:
        print(("Error multiprocesses"))

But if you are having trouble even passing a lock to your function, and you haven't bothered reading the docs then I would suggest you restructure your code so that run_smartphone only reads from the database (shouldn't require a lock) and you write to the db in the for loop (doesn't require a lock). Otherwise, you're going to get yourself into a deadlock quagmire.