Re: [globus-python] keyboard interrupt?

[Keith Jackson <krjackson@lbl.gov>]

Edinburgh perhaps for the beer?
cheers,
--keith

Jonathan P Giddy wrote:

> Quoting Keith Jackson <krjackson@lbl.gov>:
> 
> 
>>Yes, and I haven't figured out how to fix it yet. If anyone has any good 
>>ideas, I'd buy them a beer at the next conference. :-)
>>--keith
>>
>>
> 
> OK, then, this happens because the Python interpreter only allows the 
> "main" thread to handle signals (and hence Ctrl-C).  When a signal is 
> caught, it is put in a queue, and every time round the bytecode evaluation 
> loop the main thread checks for any pending signals.  If it's a SIGINT, a 
> KeyboardInterrupt is raised.
> 
> When the main thread is waiting in a system call, it doesn't check the 
> queue until the call finishes and it evaluates more bytecodes. An accept()
> of course doesn't return until a network connection occurs.
> 
> Two possible solutions:
> 
> 1. Put all your code in non-main threads, and create a simple main thread
> that does "while not done: time.sleep(0.2)", ensuring the thread handles
> a signal in 0.1 of a second on average.
> 
> 2. Put the above timeout in a select:
> while not done:
>     r = select.select([sock], [], [], 0.2)
>     # any pending signals handled now
>     if sock in r:
>         conn, addr = sock.accept()
>         ...
> 
> Cheers,
>     Jon.
> 
> 
> 
>>Robert Olson wrote:
>>
>>
>>>Does anyone else see the effect where a GSISocketServer listener cannot 
>>>be interrupted with the keyboard interrupt until a network request 
>>>actually comes in?
>>>
>>>--bob
>>>
>>>
> 
> --
> Jonathan Giddy
> Grid Technologies Co-ordinator
> The Welsh e-Science Centre
> Cardiff University
> ph +44 (0)29 2087 6998
> fax +44 (0)29 2087 4598
> J.P.Giddy@wesc.ac.uk
> 
> 


-- 
----------------------------------------------------------------
Keith R. Jackson                        KRJackson@lbl.gov
Grid Technology Group			(510) 486-4401
Lawrence Berkeley National Laboratory  	http://www-itg.lbl.gov/~kjackson/