glued bdb databases limit ?

["Raphaël 'SurcouF' Bordet" <surcouf@debianfr.net>]

Hi, 

I've a slapd instance which have more than 28 databases glued using
OpenLDAP 2.3.35/BerkeleyDB 4.2.52. Most of theses databases (back_bdb)
doesn't have memory cache setting greater than 1 Mb using DB_CONFIG.
Greaters have 5 Mb.
The slapd process consums about 1454 Mb of RAM.
But with this following search, I'm getting a curious error message : 

$ ldapsearch -x -LLL -H ldapi:// -D 'cn=admin,dc=debianfr,dc=net' -W -b
'dc=debianfr,dc=net' -s one '(objectClass)' dn 
[...]
dn: ou=28th,dc=debianfr,dc=net
Additionnal information: internal error

I don't get no more data after this error message.
And in log file of slapd, I've theses lines : 

/----( slapd.log )----
Jun 22 15:45:01 meta2 meta-slapd[5676]: [ID 215403 local4.debug] conn=98
op=1 BIND dn="cn=manager,dc=defense,dc=gouv,dc=fr" method=128
Jun 22 15:45:01 meta2 meta-slapd[5676]: [ID 600343 local4.debug] conn=98
op=1 BIND dn="cn=manager,dc=defense,dc=gouv,dc=fr" mech=SIMPLE ssf=0
Jun 22 15:45:01 meta2 meta-slapd[5676]: [ID 618536 local4.debug]
do_abandon: bad msgid 0
Jun 22 15:45:01 meta2 meta-slapd[5676]: [ID 588225 local4.debug] conn=98
op=1 RESULT tag=97 err=0 text=
Jun 22 15:45:01 meta2 meta-slapd[5676]: [ID 469902 local4.debug] conn=98
op=2 SRCH base="dc=debianfr,dc=net" scope=2 deref=0
filter="(objectClass=*)"
Jun 22 15:45:01 meta2 meta-slapd[5676]: [ID 744844 local4.debug] conn=98
op=2 SRCH attr=dn
Jun 22 15:45:05 meta2 igc-slapd[5674]: [ID 887893 local4.debug]
do_syncrep2: rid 001 LDAP_RES_SEARCH_RESULT
Jun 22 15:45:27 meta2 meta-slapd[5676]: [ID 631200 local4.debug]
bdb_locker_id: err Not enough space(12)
\---8<---8<---8<---8<---8<---

I've already looking for an database problem but I can disable any
database, when I'm adding the 29th database, this error come back !

Any ideas ?

Regards,

-- 
RaphaÃl 'SurcouF' Bordet