Re: Startup time

[Buchan Milne <bgmilne@staff.telkomsa.net>]

On Tuesday 11 September 2007 23:54:14 Sumith Narayanan wrote:
> Thanks Everyone  !
>
> Here is my system configuration :
>
> OpenLDAP version : 2.3.27
> Berkeley DB Version: 4.4.20
> Server : Mac OSX Tiger in Power PC
> Memory : 8 GB
> Number of physical DB : 3
> Size of each DB : 4 , 12 and 24 each.
>
> DB_CONFIG file for one DB , others are also set simillarly :
>
> ==========
> # $OpenLDAP: pkg/ldap/servers/slapd/DB_CONFIG,v 1.1.2.3 2006/08/17
> 17:36:19 kurt Exp $
> # Example DB_CONFIG file for use with slapd(8) BDB/HDB databases.
> #
> # See Sleepycat Berkeley DB documentation
> #   <http://www.sleepycat.com/docs/ref/env/db_config.html>
> # for detail description of DB_CONFIG syntax and semantics.
> #
> # Hints can also be found in the OpenLDAP Software FAQ
> #       <http://www.openldap.org/faq/index.cgi?file=2>
> # in particular:
> #   <http://www.openldap.org/faq/index.cgi?file=1075>
>
> # Note: most DB_CONFIG settings will take effect only upon rebuilding
> # the DB environment.
>
> # one 0.20 GB cache
> set_cachesize   0       42428800        0

You should probably assign more BDB cache than this (and reduce your entry 
cache size).

> # Data Directory
> #set_data_dir db
>
> #set db flags
> #only use when using slapdADD
> set_flags   DB_TXN_NOSYNC

Don't do this, rather run 2.3 and use the -q flag.

> set_lk_max_locks         2000
>
> # Transaction Log settings
> set_lg_regionmax        1048576
> set_lg_max              20485760
> set_lg_bsize            2097152
> set_lg_dir
> /Volumes/ngs/app/ldapp/openldap/var/openldap-data/db/externals/
>
> # Automatically remove log files that are no longer needed.
> set_flags DB_LOG_AUTOREMOVE

Which means, you should usually only have one transaction log file. How many 
do you have?

> # Note: special DB_CONFIG flags are no longer needed for "quick"
> # slapadd(8) or slapindex(8) access (see their -q option).
> ==========
>
> slapd.conf file :
> ---------------------
>
> ====
> # Do not enable referrals until AFTER you have a working directory
> #
> # See slapd.conf(5) for details on configuration options.
> # This file should NOT be world readable.
> #
> include        
> /Volumes/ngs/app/ldapp/openldap/etc/openldap/schema/core.schema include
> /Volumes/ngs/app/ldapp/openldap/etc/openldap/schema/cosine.schema
> include
> /Volumes/ngs/app/ldapp/openldap/etc/openldap/schema/inetorgperson.schema
> # Custom Schema
> include        
> /Volumes/ngs/app/ldapp/openldap/etc/openldap/schema/ist.schema # Define
> global ACLs to disable default read access.
>
> # service AND an understanding of referrals.
> #referral       ldap://root.openldap.org
>
> pidfile         /Volumes/ngs/app/ldapp/openldap//var/run/slapd.pid
> argsfile        /Volumes/ngs/app/ldapp/openldap//var/run/slapd.args
> replogfile
> /Volumes/ngs/app/ldapp/openldap/var/openldap-slurp/replogfile.log
>
> # Give the replicator account the ability to update and everyone read
> access.
>
> access to attrs=userpassword
>        by * auth
> access to *
>        by dn.base="cn=replicator,o= Computer" write
>        by * read
>
> access to * by * write
>             by dn.base="cn=manager,o= computer" write
> readonly off
>
> loglevel       256
> #conn_max_pending 300
> defaultsearchbase "o= Computer"
> gentlehup          on
> idletimeout        300
> sizelimit          2000000
> timelimit          300
> password-hash     {SSHA}
> allow bind_v2
> threads           32
>
> database        bdb
> suffix          "ou=externals,o= Computer"
> subordinate     "o= Computer"
> rootdn "cn=Manager,o= Computer"
> dbcachesize     10000000
> cachesize        100000000

There's no way you have enough memory for this (you'd probably need almost 
1TB). Please read the tuning entry in the FAQ. If you don't have enough 
memory for your entry cache, it's better to assign more to the BDB backend 
cache. You don't say how many entries you have (or what the size of 
id2entry.bdb is), but your performance is most likely going to be very poor 
with this setting, as you'll start swapping quite soon ...

> directory      
> /Volumes/ngs/app/ldapp/openldap/var/openldap-data/db/externals index       
>    objectClass      eq
> index               dsid                  eq
> index           cn              eq,sub
> index           givenName       eq,sub
> index           mail            eq,sub
> index           sn              eq,sub
> index           telephonenumber eq,sub
> index           entryUUID       eq
> replica         uri=ldap://ldapws1.corp.computer.com:3893/
>                 binddn="cn=Replicator,o= Computer"
>                 bindmethod=simple
>                 credentials=******

As I suspected, you have no checkpoint statement. So, if database recovery is 
required, it will be going through *all* the transaction log files.

> ============
>
> We are planning to upgrade to 64 bit Mac Intel processors, but till
> then we need this to be up and running.

I'd rather run Linux on commodity hardware (anything recent can run 64bit) ... 
you have a lot of expensive ram that is more or less useless to you ...

> Any suggesstions for an optimal configuration will be appreciated.

You need to provide a little bit more information (number of entries, or sizes 
of the id2entry.bdb files). But, drop your entry cache (cachesize in 
slapd.conf), increase your BDB backend cache (see the tuning entry for some 
guidelines), but ensure you don't assign more cache than you have the memory 
for ..

Regards,
Buchan