LDAP 2.0.12 Failures

[Ray Spinhirne <rays@admin.stedwards.edu>]

I have openldap 2.0.12 install on an HPUX 11.00 system.  It seemed to work
ok in testing but when put into production mode we are getting ldap failures
at the rate of about 1 per day.  

Sometimes ldap simply dies and a restart of it is ok.  

At other time the database seems to be corrupted.

    1.  Somethimes a slapindex will fix it.
    2.  Somethimes it only show a few of the records and we must restore
        from backup.
    3.  This last time the index build with a -v runs "forever" and 
        shows continually loops thru about 20 records.

Needless to say this is not a good situation when we are trying to use
the as the primary authentication method for all facutly, staff, and 
students.   

I have download 2.0.15 (but not attempted to install it).  

Does anyone know how to fix these problems or will I be forced to go back
to a Netscape (Iplanet) solution.

I am using version 3.3.11 of Berkeley DB from Sleepycat Software as
the database.

Messages like the following in syslog are the only error indications.
However, these message occur without any obvious problems in the server.

Oct  3 09:14:46 newl syslog: ldbm: ==> set_alloc: method meaningless after open
Oct  3 09:14:47 newl syslog: ldbm: ==> set_alloc: method meaningless after open
Oct  3 09:19:23 newl  above message repeats 3 times
Oct  3 09:26:26 newl syslog: ldbm: ==> set_alloc: method meaningless after open
Oct  3 09:27:38 newl syslog: ldbm: ==> set_alloc: method meaningless after open
Oct  3 09:33:43 newl  above message repeats 7 times
Oct  3 09:44:24 newl syslog: ldbm: ==> set_alloc: method meaningless after open
Oct  3 09:44:51 newl slapd[26020]: ldbm: ==> set_alloc: method meaningless 

Thanks

Ray Spinhirne
Director Computer Services
St. Edward's University
Austin TX