free the workers
free the workers
Currently the workers for parallel processing aren't closed properly.
Details
Currently, we only use the .close()
statement like
pool = multiprocessing.Pool(n_process)
pool.apply_async(...)
pool.close()
but as stated in the official docsthe
.close()` statement only prevents the pool to get further task submissions. It will not close the pool itself.
Tasks
-
add pool.join()
after allpool.close()
calls to exit the pool properly.-
data handlers -
preprocessing -
postprocessing -
plotting
-
Edited by Ghost User