Re: LDAP hangs requiring slapd stop and deleting __db.00* files

[Howard Chu <hyc@symas.com>]

Owen DeLong wrote:

> I'm seeing this problem also, using openldap version 2.2.23.  In my case,
> clearing things up requires the following steps:
>
> service ldap stop
> /usr/sbin/slapd_db_recover -h /var/lib/ldap
> service ldap start
> chown -R ldap.ldap /var/lib/ldap
> service ldap start
>
> For some reason, the db_recover deletes:
> -rw-------   1 ldap ldap   16384 Mar 28 22:22 __db.001
> -rw-------   1 ldap ldap  278528 Mar 28 22:22 __db.002
> -rw-------   1 ldap ldap   98304 Mar 28 22:22 __db.003
> -rw-------   1 ldap ldap  450560 Mar 28 22:22 __db.004
> -rw-------   1 ldap ldap   24576 Mar 28 22:22 __db.005
>
> and the first service ldap start creates them owned by root, then, ldap
> dies because they are owned by root and not ldap.  The second service
> ldap start starts the ldap daemon.
>  
As the OpenLDAP project does not provide start/stop scripts, this is an 
issue for you to take up with whoever packaged the version you're using, 
or whoever wrote your service start script.

> I don't know about idlcachesize... What is a reasonable value to put there?
> It is not documented or even mentioned in my slapd.conf(5) manpage.
> Is it new to 2.2.24?
>  
The idlcachesize setting was introduced in 2.1.10 and has been in the 
slapd-bdb(5) manpage ever since.

--
 -- Howard Chu
 Chief Architect, Symas Corp.       Director, Highland Sun
 http://www.symas.com               http://highlandsun.com/hyc
 Symas: Premier OpenSource Development and Support