[Date Prev][Date Next]
[Chronological]
[Thread]
[Top]
Re: OpenLDAP and BIG mail environments
--On Thursday, February 12, 2004 12:32 AM -0600 Micah Anderson
<micah@riseup.net> wrote:
Quanah,
Can you give some details of your slapd.conf and any tunables you've
done in your ldap.conf or anything you've done to tune your BDB?
From slapd.conf:
---------------------------------------------------------------------------
# Schema check allows for forcing entries to match schemas for
# their objectClasses's
schemacheck on
# Read slapd.conf(5) for possible values
loglevel 256
# Set the number of threads
threads 20
# Set the idletimeout for threads
idletimeout 30
#######################################################################
# bdb database definitions
#######################################################################
database bdb
suffix "dc=stanford,dc=edu"
rootdn "cn=Manager,dc=stanford,dc=edu"
# Cleartext passwords, especially for the rootdn, should
# be avoid. See slappasswd(8) and slapd.conf(5) for details.
# Use of strong authentication encouraged.
#rootpw secret
# The database directory MUST exist prior to running slapd AND
# should only be accessible by the slapd/tools. Mode 700 recommended.
directory /db
# Checkpointing
checkpoint 1024 5
# Entries to cache in memory
cachesize 10000
---------------------------------------------------------------------------
From /db/DB_CONFIG
---------------------------------------------------------------------------
# $Id: DB_CONFIG,v 1.7 2004/01/28 19:10:58 quanah Exp $
set_cachesize 2 0 1
set_lg_regionmax 262144
set_lg_bsize 2097152
set_lg_dir /var/log/bdb
# Just use this setting when doing slapadd...
#set_flags DB_TXN_NOSYNC
#set_flags DB_TXN_NOT_DURABLE
---------------------------------------------------------------------------
We've got schemacheck off and our threads increased to 64, cachesize
2000 and not much else that we've been tweaking, you can see our
graphs culled from the monitor stuff here:
http://risuep.net/mrtg/buffy_slapd
In my testing, I found that increasing the threads past a certain point
*decreased* performance. We found 20 to be the best number of threads for
our configuration. Our cachesize is 5 times your, however.
Does it make any sense to move our DB
files form the /var partition since they are probably all cached in
memory anyways?
The main thing in 2.1 to think about, disk wise, is that your BDB log files
should be kept on a separate disk from your DB itself. We have 2 drives,
and /db is the only partition on the second drive. /var/log/bdb is located
on the first drive.
Also note that we are running Solaris 8 with 4 GB of RAM and not Linux. ;)
I have 2 new linux servers (Dell 1750's, 2 3GHz CPU's, 4GB memory) set up
to test soon.
One other question -- Do you have a local DNS caching service running on
your linux box? We turn on reverse lookups in our slapd.conf, so we also
installed a local DNS cacher.
--Quanah
--
Quanah Gibson-Mount
Principal Software Developer
ITSS/TSS/Computing Systems
ITSS/TSS/Infrastructure Operations
Stanford University
GnuPG Public Key: http://www.stanford.edu/~quanah/pgp.html