Re: [sup-talk] Problems running sup-convert-ferret-index

[Steve Goldman <sgoldman@tower-research.com>]

Excerpts from Rich Lane's message of Sun Jan 17 21:10:17 -0500 2010:
> Excerpts from Steve Goldman's message of 2010-01-17 20:44:11 -0500:
> > Excerpts from Steve Goldman's message of Sun Jan 17 19:28:12 -0500 2010:
> > > Excerpts from William Morgan's message of Sun Jan 17 19:04:46 -0500 2010:
> > > > Reformatted excerpts from Steve Goldman's message of 2010-01-17:
> > > > > ... until it exited thirty minutes later with seg fault.
> > > > > 
> > > > > Anyone else see this?  Any way to debug it?
> > > > 
> > > > Is it possible it's running out of memory?
> > > 
> > > Possible.  I have 4GB and about 60k messages in my index.  So it's not
> > > too likely.  I'll run it again and watch the usage.
> > 
> > I ran it in strace.  It hung on a call to munmap, and then crashed
> > with a lot of a bunch of system call timeout warnings.
> > 
> > So it has something to do with memory.
> > 
> > Would the ruby debugger help narrow it down?  Does the ruby debugger
> > exist?
> 
> Try compiling Xapian from source with --enable-assertions. It would also
> be useful to get a core dump (ulimit -c unlimited).

This feels like a wild goose chase.

Can I start a xapian index from scratch?  Any downside to this besides
the time sink?

Thanks.
-- 

Steve Goldman
sgoldman@tower-research.com

T: 212.219.6014
F: 212.219.6007

Tower Research Capital, LLC
377 Broadway, 11th Fl.
New York, NY 10013
_______________________________________________
sup-talk mailing list
sup-talk@rubyforge.org
http://rubyforge.org/mailman/listinfo/sup-talk