On Mon, 2005-02-21 at 18:27, Quanah Gibson-Mount wrote: > --On Monday, February 21, 2005 5:50 PM -0500 Samuel Tran <stran@amnh.org> > wrote: > > > > Okay. > > > > I guess I'll go back to BDB 4.2.52. > > > > I am readying my new LDAP cluster for production. I was using BDB 4.2.52 > > but upgraded to 4.3.27 when it was out. > > > > I am using OL 2.2.23 + BDB 4.3.27 on Linux Debian Woody. The LDAP slaves > > are 2 HP DL360 with 2 1.4Ghz Xeon CPUs and 2 Gb of RAM. The clustering > > software is Keepalived 1.1.10. > > > > I've been able to achieve about 350 LDAP queries/sec on a single slave > > with loglevel 256. > > > > Is the kind of result I should get with such a setup? > > I would say no. > > See: > > <http://www.symas.com/10k-solvlinux.shtml> > > with a reference to the machine types at > > <http://www.symas.com/benchmark.shtml> > > I would verify that your slapd.conf is properly tuned with cachesize, > idlcachesize, and the appropriate indices. I would also verify that your > DB_CONFIG file is well tuned. > Here a extract of my slapd.conf: cachesize 100000 checkpoint 512 720 index objectClass,uid,uidNumber,gidNumber,memberUid eq index cn,mail,surname,givenname eq,subinitial Here is the content of my DB_CONFIG: set_cachesize 0 52428800 0 set_lg_regionmax 1048576 set_lg_max 10485760 set_lg_bsize 2097152 set_lg_dir /var/log/openldap_bdb set_tmp_dir /tmp We are starting with less than 4,000 entries in our LDAP directory. Sam
Attachment:
signature.asc
Description: This is a digitally signed message part