Re: svn export core dumps

Peter Burkholder <peterb@ucar.edu> writes:
> The initial import of the tree worked but slowed to a crawl (maybe one or
> two adds second) while import the "broad" collection.

Hmmm. How long (how many files) did it take to slow to a crawl? Or
was it slow from the start?

> The command
> svn export http://svn.dpc.ucar.edu/records/trunk/adn
> chugged along until:
>
> A adn/sercnagt/oai%3Aserc.carleton.edu%3ASERC-NAGT-000-000-000-278.xml
> A adn/sercnagt/oai%3Aserc.carleton.edu%3ASERC-NAGT-000-000-000-279.xml
> A adn/broad
> Segmentation fault (core dumped)
>
> The export was stuck on the last Add for about 30 minutes before seg
> faulting.

Ooooh. That's not good.

Can you get a backtrace?

> Vital stats:
>
> Both server and client:
> BerkeleyDB 4.1

Have you tried with BDB 4.0.14? That's the max we officially support
right now (heh, not that there's anything "official" about our support
in any case).

> Linux Redhat 7.1
> svn 0.24.2
> server
> httpd 2.0.46
> client
> httpd built against 2.1 HEAD (slightly dated)

The svn client doesn't use any httpd code.

> Is this a bug that needs fixing or am I doing something wrong? Let me know
> if I can provide more info or tests.

They're certainly bugs (both the import slowness, and the export
segfault), and yes they need fixing :-).

Is the repository available on the web? Can we try the export
ourselves?

Your data set is a good test for us; please be patient while we figure
out what's happening, it'll pay off in the long run.

Thanks,
-Karl

---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Received on Fri Jul 4 06:38:16 2003