Hello all, new to this forum (member of many others). Hopefully I can find help here.
SERVER:
Brand new server Oracle Enterprise SPARC T4-1
Loaded Solaris SPARC 10 u10, patched to 147440-27
Loaded OpenLDAP v2.4.30
Loaded Berkley DB 4.7.25.NC Loaded OpenSSL 1.0.1c
Note: All packages are from sunfreeware
For sake of simplicity:
/ldap/openLdapData
/ldap/openLdapLogs
In the logs directory there are 50m files that are being created every min, and filling up the 100GB of space on that filesystem. The file system is from a ZFS pool with a ZFS quota of 100GB (if any of that matters to anyone). I tried lowering the default logging level from 256 to 128+32+8=168 by placing entries in the slapd.conf file. Should those entries also be in the DB_CONFIG file as well. Also in the DB_CONFIG file is "set_flags DB_LOG_AUTOREMOVE", but it is still doing it. I am tempted to go to 0, but I don't want to have to do that.
I have this same OS and software running on some T2000, T51xx, and T52xx series servers with no issues. I have configured hundreds of servers using these packages, and their predecessors. I have never seen this behavior before. There seems to be a lot of things different about the T4-1 that have made configuring this server a giant PITA.
Does anyone know what the issue is with this thing? 
What are these 50m files created every minute (do you really mean 50 million files created every minute or 50 MB of new logs every minute ?)
In any case, why don't you post a sample of these logs ?
Are you doing snapshots of the file system containing the logs ?
Thanks for the reply.
These are log files created automatically for the purpose of auto-recovering the LDAP from a failure. In most systems I have, there are just 2 - 6 files (50mb each) in the logs dir. These files are 50mb due to the "Transaction Log Settings" in the DB_CONFIG. Increasing / decreasing the configured size has no affect on the problem. These files normally "track" changes made to LDAP DB via the Berkeley DB (BDB). The problem is, these files are not human readable, and it is on an isolated system not connected to the internet (IE: we are not allowed to post files). These log files are "usually" automatically cleaned up by a statement in the conf file:
set_flags DB_LOG_AUTOREMOVE
This setting works on every other system we have, with the exception of this T4-1.
No, there are no snapshots being done. It is on a raidz2 zfs file-system, but that shouldn't matter..
It has been suggested by some folks @ OpenLDAP that the older version of BDB -v4.7 is the issue. I have to claim partial ignorance to that statement since we don't use BDB for anything, but to support OpenLDAP.