[prev in list] [next in list] [prev in thread] [next in thread]
List: openldap-software
Subject: Re: seg faults with 2.12 slapd and GSSAPI bind
From: Stephen Frost <sfrost () snowman ! net>
Date: 2003-02-28 14:03:18
[Download RAW message or body]
* Ben Poliakoff (benp@reed.edu) wrote:
> * Howard Chu <hyc@highlandsun.com> [030227 15:37]:
> > Time and again we advise that the MIT Kerberos 5 libraries are not usable
> > with SASL and OpenLDAP. You should not be surprised that ignoring that advice
> > causes problems. Nor should you have to scratch your head too hard thinking
> > about any solutions. Use Heimdal.
>
> OK, I'm already working on a build of heimdal and a rebuild of sasl.
Be sure to check out the post to this list earlier which explains all
that you have to do to get Heimdal to compile as threadsafe. As I
recall you have to use a CVS version after a certain date and set an
environment variable or pass something on the command-line to build it.
> In retrospect I did spot:
>
> How can OpenLDAP, Cyrus-SASL, OpenSSL and KerberosV be combined?
> http://www.openldap.org/faq/data/cache/544.html
>
> For what it's worth. Compiling OpenLDAP with and without threads made
> no difference.
That would indicate to me, at least, that the problem (unsuprisingly) is
not the constantly blamed MIT krb5 libs (which have worked fine for me
so far, though I do believe that under load they can cause problems). I
believe someone else is working on a patch, or at least looking into the
possibility of doing something, to deal with the MIT krb5 libs not being
thread-safe.
> I'll shut up now.
If you have the time to I'd be happy to work through this issue with
you. I'll go back through the mailing list and see if I can figure out
your setup and duplicate what you're getting. I don't think I saw a
backtrace with debugging symbols, do you think you could do one?
Stephen
[Attachment #3 (application/pgp-signature)]
[prev in list] [next in list] [prev in thread] [next in thread]
Configure |
About |
News |
Add a list |
Sponsored by KoreLogic