Security Software

Hello,

We are running a pair of master-slave LDAP servers (Sun One Directory Server 5.2 SP4) on Sun hardware with enough disk and memory resources. The database has more than 1.4M records, the servers have 4GB RAM installed and more than 1GB of RAM is free; each server is not loaded more then 10% of CPU.

Sometimes the following error messages are appearing in LDAP's error log-file on either of the two servers (not at the same time):

[23/Jun/2006:11:34:16 +0300] - DEBUG - conn=-1 op=-1 msgId=-1 - database index operation failed BAD 1050, err=12 Not enough space
[23/Jun/2006:11:34:16 +0300] - DEBUG - conn=-1 op=-1 msgId=-1 - libdb: Unable to allocate 8311 bytes from mpool shared region: Not enough space

Sometimes after such error messages the complaining server stops to delete old transaction log-files ("log.*" files in /db directory) and the files are starting to flood the disk; the server also fails to perform grace shutdown of "ns-slapd" daemon.

In accordance with the recommendations which I've found in the forum a database backup-restore operation was performed with no luck - the servers still sometimes complain about the same problem.

I have some feeling that the problem is somehow connected with database index tuning and that the servers are running out of their internal index space but I can not find anywhere in Sun One Directory Server documenation any reference about tuning of memory allocation for indexes.

Any ideas?

Kind regards,
Victor