Hi, there is an issue driving me crazy with the web2py scheduler:
If you return something that has a huge size then it will always timeout;
even if the scheduler task correctly finishes. Let me explain with an
example:
def small_test():
s = 's'*1256018
another_s = s
#print s
#print another_s
#print 'FINISHED PROCESS'
return dict(s = s, another_s = another_s, f = 'finished')
small_test is the function to execute, as you can see a string full of 's'
1256018 times is. Simple
So when you enqueue the scheduler every time the output is the same:
http://prnt.sc/a9iarj (screenshot of the TIMEOUT)
As you can see from the screenshot, the process actually finished; while
logging the scheduler output shows the following:
DEBUG:web2py.scheduler.PRTALONENETLAPP-SRV#24475: work to do 1405
DEBUG:web2py.scheduler.PRTALONENETLAPP-SRV#24475: new scheduler_run
record
INFO:web2py.scheduler.PRTALONENETLAPP-SRV#24475:new task 1405 "small_test"
portal/default.small_test
DEBUG:web2py.scheduler.PRTALONENETLAPP-SRV#24475: new task allocated:
portal/default.small_test
DEBUG:web2py.scheduler.PRTALONENETLAPP-SRV#24475: task starting
DEBUG:web2py.scheduler.PRTALONENETLAPP-SRV#24475: task started
DEBUG:web2py.scheduler.PRTALONENETLAPP-SRV#24475: new task report:
COMPLETED
DEBUG:web2py.scheduler.PRTALONENETLAPP-SRV#24475: result: {"s":
"ssssssssssssssssssssssssssssssssssssssssssssssssssssssssssssssssssssssssssssssssss$
DEBUG:web2py.scheduler.PRTALONENETLAPP-SRV#24475:........recording
heartbeat (RUNNING)
DEBUG:web2py.scheduler.PRTALONENETLAPP-SRV#24475:........recording
heartbeat (RUNNING)
DEBUG:web2py.scheduler.PRTALONENETLAPP-SRV#24475:........recording
heartbeat (RUNNING)
DEBUG:web2py.scheduler.PRTALONENETLAPP-SRV#24475:........recording
heartbeat (RUNNING)
DEBUG:web2py.scheduler.PRTALONENETLAPP-SRV#24475:........recording
heartbeat (RUNNING)
DEBUG:web2py.scheduler.PRTALONENETLAPP-SRV#24475: freeing workers that
have not sent heartbeat
DEBUG:web2py.scheduler.PRTALONENETLAPP-SRV#24475:........recording
heartbeat (RUNNING)
DEBUG:web2py.scheduler.PRTALONENETLAPP-SRV#24475:........recording
heartbeat (RUNNING)
DEBUG:web2py.scheduler.PRTALONENETLAPP-SRV#24475:........recording
heartbeat (RUNNING)
DEBUG:web2py.scheduler.PRTALONENETLAPP-SRV#24475:........recording
heartbeat (RUNNING)
DEBUG:web2py.scheduler.PRTALONENETLAPP-SRV#24475:........recording
heartbeat (RUNNING)
DEBUG:web2py.scheduler.PRTALONENETLAPP-SRV#24475: freeing workers that
have not sent heartbeat
DEBUG:web2py.scheduler.PRTALONENETLAPP-SRV#24475:........recording
heartbeat (RUNNING)
DEBUG:web2py.scheduler.PRTALONENETLAPP-SRV#24475:........recording
heartbeat (RUNNING)
DEBUG:web2py.scheduler.PRTALONENETLAPP-SRV#24475:........recording
heartbeat (RUNNING)
DEBUG:web2py.scheduler.PRTALONENETLAPP-SRV#24475:........recording
heartbeat (RUNNING)
DEBUG:web2py.scheduler.PRTALONENETLAPP-SRV#24475:........recording
heartbeat (RUNNING)
DEBUG:web2py.scheduler.PRTALONENETLAPP-SRV#24475: freeing workers that
have not sent heartbeat
DEBUG:web2py.scheduler.PRTALONENETLAPP-SRV#24475:........recording
heartbeat (RUNNING)
DEBUG:web2py.scheduler.PRTALONENETLAPP-SRV#24475: task timeout
DEBUG:web2py.scheduler.PRTALONENETLAPP-SRV#24475: recording task report in
db (TIMEOUT)
DEBUG:web2py.scheduler.PRTALONENETLAPP-SRV#24475: status TIMEOUT, stop_time
2016-02-29 11:56:52.393706, run_result {"s": "sssssssssssssssssssssssssss$
INFO:web2py.scheduler.PRTALONENETLAPP-SRV#24475:task completed (TIMEOUT)
DEBUG:web2py.scheduler.PRTALONENETLAPP-SRV#24475:looping...
INFO:web2py.scheduler.PRTALONENETLAPP-SRV#24475:nothing to do
As you can see there is a TaskReport object in the queue with a COMPLETED
status (I know this because I read the scheduler.py code of web2py) So I'm
pretty sure the task finishes quite fast but then it hangs.
So I did another test, that doesn't directly use the scheduler but only
calls the executor method from the scheduler and usess process; just like
the scheduler would:
from gluon.scheduler import Task
from gluon.scheduler import executor
t = Task(app='portal', function='small_test', timeout = 120)
import logging
logging.getLogger().setLevel(logging.DEBUG)
import multiprocessing
queue = multiprocessing.Queue(maxsize = 1)
out = multiprocessing.Queue()
t.task_id = 123
t.uuid = 'asdfasdf'
p = multiprocessing.Process(target=executor, args=(queue, t, out))
p.start()
p.join(timeout = 120)
p.is_alive()
when the join finishes waiting (2 minutes) if you check for p.is_alive() it
always returns True; but when you do a queue.get() and then instantly check
for p.is_alive() the process finishes!!!!!
So i noticed the problem is from multiprocessing library, due to the fact
that it can't handle lots of data from a queue (which seems kind of strange
for my case, but I don't know how it is implemented); anyways i found this
bug: http://bugs.python.org/issue8237 and http://bugs.python.org/issue8426
The interesting part is it is actually documented (I didn't knew that):
https://docs.python.org/2/library/multiprocessing.html#multiprocessing-programming
But in my current implementation this will happen quite often, I'll work on
a work-around but I would really like that web2py scheduler could handle
large data output from my processes for me, but well that is my wish and I
would like to have some guidance on this issue and avoid a work-around.
Anyway, this should be documented somewhere in web2py too (that probably
could had saved me a week of code reading and debugging); or it should be
managed somehow (I wouldn't naturally expect an output limit besides the
database implementation).
--
Resources:
- http://web2py.com
- http://web2py.com/book (Documentation)
- http://github.com/web2py/web2py (Source code)
- https://code.google.com/p/web2py/issues/list (Report Issues)
---
You received this message because you are subscribed to the Google Groups
"web2py-users" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to web2py+unsubscr...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.
