Re: (ITS#3648) ldbm hangs during shutdown

[hyc@symas.com]

Interesting. Probably the best way to fix this is to change this code to 
use the slapd runqueue instead of spawning its own thread.

vek@pharmapartners.nl wrote:

>Full_Name: Villy Kruse
>Version: 2.2.24
>OS: Linux NPTL
>URL: ftp://ftp.openldap.org/incoming/villy-kruse-050411.patch
>Submission from: (NULL) (62.58.72.98)
>
>
>
>The ldbm backend, or rather the cache sync daemon thread, may deadlock
>during shut down.  The problem started when installing openldap on a
>RHAS3 with NPTL threads library.  When this deadlock occurs only kill -9
>can terminate slapd.  The problem seems not to occur on RH7.3 with
>linux threads.
>
>The cause of the problem is that the sync daemon thread will access
>the connection_pool structure and its mutex after this structure has
>been destroyed.
>
>The included patch prevents this from happening -- most of the time.
>
>openldap version 2.2.24
>BerkeleyDB version 4.3.27 + 3 SleapyCat patches.
>RedHat glibc-2.2.5-44 with NPTL
>RedHat Linux kernel 2.4.21
>Tripple Intel Xeon CPU in an IBM server
>
>
>  
>


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