Re: Back to OpenLDAP 1.3.6

["Kurt D. Zeilenga" <Kurt@OpenLDAP.org>]

At 02:39 PM 8/25/99 +0200, Frédéric Poels wrote:
>Good afternoon!
>
>I've had the bad (or good?) idea to change some parameters of a
>multi-threaded OpenLDAP 1.3.6 slapd server running on AIX 4.2.1:
>cachesize=100
>dbcachesize=200

200 bytes isn't much of a dbcachesize.  In fact, with GDBM
it translates to zero.  But it *should* work...

>The directory contains about 15000 entries. I perform the following
>operations:
>- A process forks 15 times, each son binds to the LDAP server and perform a
>different search which returns about 1000 entries. The father waits for its
>sons to terminate and then exits.
>- It works pretty well :o)
>- I do it again
>- All sons return "No such object" :o(
>- Any request sent to the LDAP server fail with the "No such object" reason
>code :o/
>
>Have I missed something? Or is it a "known bug"?

You have likely exhausted some resource.  slapd is known not
to react well in such situtations.

Kurt