RE: Slapd fails to die (ITS#2502)

[hyc@highlandsun.com]

> -----Original Message-----
> From: owner-openldap-bugs@OpenLDAP.org
> [mailto:owner-openldap-bugs@OpenLDAP.org]On Behalf Of quanah@stanford.edu

> --On Friday, May 09, 2003 6:39 AM -0700 "Kurt D. Zeilenga"
> <Kurt@OpenLDAP.org> wrote:
>
> > You might try connecting gdb to the running process and
> > seeing what each thread is doing.
>
> Kurt,
>
> Thanks! It looks like from the output, that with thread 10,
> it is holding
> on to a modification?
> Let me know if there is anything more that may be helpful in
> debugging this.

The lock info shows that the modify op was looping because it was waiting for
a read lock on the entry cache to be released. Since there were no other
active threads, there should not have been any outstanding read locks.
Unfortunately there's not enough info remaining in this core image to
determine which operation left its read lock lying around. I have committed a
fix to CVS HEAD to force the release of any remaining locks before a thread
exits. This is really just a work-around, until we can find the actual bug.

  -- Howard Chu
  Chief Architect, Symas Corp.       Director, Highland Sun
  http://www.symas.com               http://highlandsun.com/hyc
  Symas: Premier OpenSource Development and Support