timing issues

["Stacy Rally" <srally@wgate.com>]

I have several questions about LDAP:

1. If you want to change the cachesize / dbcachesize after starting up slapd
and loading up the database, what is the best way to go about this.  Does
killing slapd and restarting it pick up the changes?  Do you need to
regenerate the data?

2. I was trying to add an index to my database, so I ran ldbmcat, and
created an ldif file.  I then ran ldif2index, and I get the following error:

/usr/local/etc/openldap/schema/core.schema: line 114: expecting "requires"
or "allows" got "2.5.20.1"
<oc clause> ::= objectclass <ocname>
                [ requires <attrlist> ]
                [ allows <attrlist> ]

Does anyone know why this is happening?  Am I doing something wrong?

3. I'm having major timing issues with my data.  I have a database with
between 250,000 & 300,000 entries.  All the entries have an objectclass of
"testuser" (made up by me).  Each entry has 18 attributes, all of them
filled in (none blank) with one value.  When I try to count entries, (either
by doing an ldapsearch command on the entire directory with a filter of
"objectclass=testuser", or by writing a piece of code in C to search with
the same filter, and count the entries after the search) it takes hours to
do the count. (In fact, it hasn't finished the count yet, because I have
timelimit set to 2 hours, and it times out at that point).  So, basically,
it takes over two hours to give me a count of entries with
objectclass=testuser when there are 250,000 entries with that objectclass.
I have an eq index on objectclass, but no other indexes (I didn't think they
would help, since they aren't part of my filter).  I have loglevel=0, and
I've played around with the cachesize and dbcachesize parameters, with no
noticable change in search time.  Are there other parameters that I should
be trying to change?  Is there anything else I can do to help this along?

Thanks,
Stacy Rally