slapd index and db errs crashing frequently (ITS#309)

[ngs@elok.com]

Full_Name: Neil Smith
Version: 1.2.7 release
OS: FreeBSD 2.2.8
URL: ftp://ftp.openldap.org/incoming/
Submission from: (NULL) (194.222.112.235)


Recently upgraded to OpenLDAP 1.2.7 and Berkeley DB 2.7.7 to get around memory
leaks.
Rebuilt the database (about 150,000 entries) from LDIF.
Now experience these errors in syslog and slapd restarting every minute or so.
Previous dn.dbb file was 17MB, now it is 299MB.

Sep 27 15:26:46 work slapd[18596]: ldbm_db_errcall(): ==>
/disk2/p2-overflow/slapd/db.new/dn.dbb: page 1347241258 doesn't exist, create
flag not set
Sep 27 15:26:54 work slapd[18596]: ldbm_db_errcall(): ==>
/disk2/p2-overflow/slapd/db.new/dn.dbb: page 1398095146 doesn't exist, create
flag not set
Sep 27 15:26:55 work slapd[18596]: ldbm_db_errcall(): ==>
/disk2/p2-overflow/slapd/db.new/dn.dbb: page 743589162 doesn't exist, create
flag not set
Sep 27 15:28:42 work slapd[18596]: ldbm_db_errcall(): ==>
/disk2/p2-overflow/slapd/db.new/rvnewusers.dbb: page 0: reference count
overflow
Sep 27 15:33:44 work slapd[18869]: ldbm_db_errcall(): ==>
/disk2/p2-overflow/slapd/db.new/rvnewusers.dbb: page 0: reference count
overflow
Sep 27 15:33:45 work slapd[18869]: ldbm_db_errcall(): ==>
/disk2/p2-overflow/slapd/db.new/rvnewusers.dbb: page 0: reference count
overflow
Sep 27 15:38:39 work slapd[19029]: ldbm_cache_open no unused db to close -
waiting 
Sep 27 15:38:39 work slapd[19029]: ldbm_cache_open no unused db to close -
waiting 
Sep 27 15:39:56 work slapd[19060]: ldbm_db_errcall(): ==>
/disk2/p2-overflow/slapd/db.new/dn.dbb: page 1027885397 doesn't exist, create
flag not set
Sep 27 15:39:56 work slapd[19060]: ldbm_db_errcall(): ==>
/disk2/p2-overflow/slapd/db.new/dn.dbb: page 1027885397 doesn't exist, create
flag not set
Sep 27 15:39:57 work slapd[19060]: ldbm_db_errcall(): ==>
/disk2/p2-overflow/slapd/db.new/dn.dbb: page 1346192426 doesn't exist, create
flag not set

slapd process can grow to up to 480MB in size

Should I change slapd or Berkeley DB?

Thanks,

Neil.