[yt-dev] Strage Behavior from Rockstar

[Matthew Turk]

Hi Stephen,

Okay, this definitely looks like it's hanging in MPI.  What is
probably happening is that the top communicator is different at
instantiation time than it should be; so when you access the -1 and
grab the grid, it think there are more objects in the top communicator
than there should be.

You can see this by looking at the top communicator's size attribute
-- my guess is that it *should* be 1, but is in fact NProc.  Can you
veryify?

-Matt

On Wed, Nov 14, 2012 at 4:09 PM, Stephen Skory <s at skory.us> wrote:
> Hi Matt,
>
> Thanks for answering. I've tried your suggestions below, but they
> didn't produce anything useful I think. Let me know if you have any
> other thoughts, please!
>
>>  SIGUSR1 to the
>> python process.  What does that report?
>
> Doing "kill -s SIGUSR1 PID" reports nothing, and in fact accomplishes
> nothing, the python process keeps going at 100%.
>
>> Alternately, where does the
>> traceback hit you when you ctrl-c out?
>
> [Warning] Network IO Failure (PID 14381): Interrupted system call
> [Network] Packet receive retry count at: 1
> [Warning] Network IO Failure (PID 14382): Interrupted system call
> [Network] Packet receive retry count at: 1
> [Warning] Network IO Failure (PID 14386): Interrupted system call
> [Network] Packet receive retry count at: 1
> --------------------------------------------------------------------------
> mpirun noticed that process rank 0 with PID 14381 on node
> casa116-239-dhcp.colorado.edu exited on signal 0 (Unknown signal: 0).
> --------------------------------------------------------------------------
> 6 total processes killed (some possibly by mpirun during cleanup)
> mpirun: clean termination accomplished
>
>
> --
> Stephen Skory
> s at skory.us
> http://stephenskory.com/
> 510.621.3687 (google voice)
> _______________________________________________
> yt-dev mailing list
> yt-dev at lists.spacepope.org
> http://lists.spacepope.org/listinfo.cgi/yt-dev-spacepope.org