Re: Checkpoint command not working in slapd.conf?

[Quanah Gibson-Mount <quanah@stanford.edu>]

--On Tuesday, January 18, 2005 11:41 AM +1100 Dave Horsfall 
<daveh@ci.com.au> wrote:

> OpenLDAP 2.2.17 on FreeBSD 4.10-STABLE (compiled from ports).
>
> I know it's not the latest, but using 2.2.19 is not an option right now.
> because it wants a new OpenSSL etc, which will cause "issues".

Actually 2.2.20 is the latest.

> Having been bitten by not detecting a corrupt (test) database, I figured
> I  may as well set things up properly, including checkpointing.
>
> Extract from slapd.conf:
>
> database        bdb
> include         /usr/local/etc/openldap/master.acl
> suffix          "dc=au,dc=company,dc=com"
> rootdn          XXX
> rootpw          XXX
> checkpoint      1024 5	<<<----------------------------
>
> directory       /usr/local/openldap-data/au
> cachesize       2000
>
> I understand that this means to checkpoint if updates have been performed
> since 1024kb or 5 minutes after the last checkpoint (and I assume this
> includes since the DB was started, as otherwise it would never checkpoint
> at all).

No, they will occur if that much activity has happened since the last write 
was received (this will be fixed in 2.3 to do what you think it does).

> DB_CONFIG:
> #
> # Size of internal cache; the format is Gb + b, no. caches.
> # 4Mb is plenty for our relatively teensy directories.
> #
> set_cachesize   0 4194304 1
>
> So, I make a trivial update, observe that the "__db" files and log have
> been updated, and twiddle my thumbs for 5 minutes.  Nope.  After 10
> minutes, I force a checkpoint with "db_checkpoint".
>
> I do it again, just to be sure.  Below, you can see the forced checkpoint
> at 10:17, and the second update at 10:19.
>
> guido:/var/db/openldap-data/au [30]# ls -lt
> total 8188
> -rw-------  1 ldap  ldap   368640 Jan 18 10:19 __db.004
> -rw-------  1 ldap  ldap  5251072 Jan 18 10:19 __db.002
> -rw-------  1 ldap  ldap    98304 Jan 18 10:19 __db.003
> -rw-------  1 ldap  ldap    24576 Jan 18 10:19 __db.005
> -rw-------  1 ldap  ldap     8192 Jan 18 10:19 __db.001
> -rw-------  1 ldap  ldap  5539588 Jan 18 10:19 log.0000000001
> -rw-------  1 ldap  ldap  1392640 Jan 18 10:17 id2entry.bdb
> -rw-------  1 ldap  ldap   327680 Jan 18 09:56 dn2id.bdb
> -rw-------  1 ldap  ldap   110592 Jan 17 13:51 objectClass.bdb
> -rw-------  1 ldap  ldap    24576 Jan 15 16:50 uid.bdb
> -rw-------  1 ldap  ldap    36864 Nov 19 12:34 cn.bdb
> -rw-------  1 ldap  ldap    24576 Nov 18 15:00 employeeNumber.bdb
> -rw-------  1 ldap  ldap    28672 Nov 18 15:00 sn.bdb
> -rw-------  1 ldap  ldap     8192 Nov 18 15:00 ciHost.bdb
> -rw-------  1 ldap  ldap    40960 Nov 18 15:00 ciApp.bdb
> lrwxrwxr-x  1 root  ldap       12 Oct 27 15:36 DB_CONFIG@ -> ../DB_CONFIG
> guido:/var/db/openldap-data/au [30]# date
> Tue Jan 18 10:59:17 EST 2005
>
> Yep; 40 minutes later, and it still hasn't checkpointed...  Googling
> around turned up nothing relevant, and I even checked the back-bdb
> sources  to make sure the option was being picked up (it was) and
> txn_checkpoint()  being called after an update (it was).
>
> What dumb thing have I done (or not done)?

Nothing, but if you read the list regularly, this issue has been discussed 
several times this month.  You need to add a checkpoint cron job or 
something similar until 2.3 is released.

--Quanah

--
Quanah Gibson-Mount
Principal Software Developer
ITSS/Shared Services
Stanford University
GnuPG Public Key: http://www.stanford.edu/~quanah/pgp.html

"These censorship operations against schools and libraries are stronger
than ever in the present religio-political climate. They often focus on
fantasy and sf books, which foster that deadly enemy to bigotry and blind
faith, the imagination." -- Ursula K. Le Guin