On Tue, Oct 25, 2011 at 9:50 AM, Jason Grout
<[email protected]> wrote:
> On 10/21/11 10:32 AM, William Stein wrote:
>
>> Instantly after doing this *demo* became much faster.
>> I noticed before doing this that there was a single sagenb server
>> Python process running in top using a lot of cpu..
>
> For future reference, it would have been *great* to get a stack trace of the
> currently executing code for the runaway process.  I followed the
> instructions in a stack overflow answer [1] and can now do, as the sagenb
> user on mod:
>
> gdb -p <PID of sagenb process>
> pystack
>
> and get a stack frame listing of a running process.
>
> As I have time, I'll look into where there might be a performance
> regression, but there have been a lot of code changes between June and now,
> and we haven't seen the performance regression on test.sagenb.org.
>
> Is there a possibility of running the new codebase for sagenb.org again and
> getting the stack trace when the process is using that much CPU?
>

I'm personally not super comfortable with this, especially if I have
to have to be the one responsible for fixing things when they
inevitably break and cause trouble.  From the second I switched back
to the old version until now, I haven't had any complaints about any
*.sagenb.org's not working, and haven't had to restart them, etc.   So
I'm only OK with this if you're 100% on deck, and I don't have to do
anything, and you switch it all back as soon as you can.

Of course, the right thing to do is write code to simulate a heavy
load, and run it against test.sagenb.org (or something like that).

 -- William

> Thanks,
>
> Jason
>
>
> [1]
> http://stackoverflow.com/questions/132058/getting-stack-trace-from-a-running-python-application/147114#147114
>
>
>
> --
> To post to this group, send email to [email protected]
> To unsubscribe from this group, send email to
> [email protected]
> For more options, visit this group at
> http://groups.google.com/group/sage-support
> URL: http://www.sagemath.org
>



-- 
William Stein
Professor of Mathematics
University of Washington
http://wstein.org

-- 
To post to this group, send email to [email protected]
To unsubscribe from this group, send email to 
[email protected]
For more options, visit this group at 
http://groups.google.com/group/sage-support
URL: http://www.sagemath.org

Reply via email to