[padb] Réf. : Re: tiny bug with--proc-summary

Ashley Pittman ashley at pittman.co.uk
Wed Dec 9 15:00:33 GMT 2009


On Wed, 2009-12-09 at 15:28 +0100, thipadin.seng-long at bull.net wrote:
> 
> Hi, 
> I have tested the patch and is OK. 

Ok, I'll commit it as is.  It's definitely a step forward as the current
code is non-deterministic.

> But I could have the main thread to be the second one from the LWP 
> (as my in previous example). 

You mean the thread with the LWP of 31719?  LWP 31718 is the one which
has consumed the most CPU cycles.  If you can suggest a good way for
padb to calculate which is the main thread then I'd be keen to hear it.

One other option might be to loop over the threads as before and report
all their values with the keys suffixed with _%d.  In the example you
gave this would give you pid=31718 pid_1=31719 pid_2=31720.  If you knew
that you wanted to see the data for the second pid you could then use
--proc-format="rank,hostname,pid_1,vmsize_1,vmrss_1,..."

As you say it's hard and I'm open to ideas on how you'd like it to work.

Ashley,

-- 

Ashley Pittman, Bath, UK.

Padb - A parallel job inspection tool for cluster computing
http://padb.pittman.org.uk





More information about the padb-devel mailing list