Re: [openldap 2.1.12 + BerkeleyDb 4.1.24] db_checkpoint pb

[Helene.Lastennet@alcatel.fr]

Matthew Hardin a écrit :

> > -----Original Message-----
> > From: owner-openldap-software@OpenLDAP.org [mailto:owner-openldap-
> > software@OpenLDAP.org] On Behalf Of Helene.Lastennet@alcatel.fr
> > Sent: Wednesday, November 19, 2003 9:48 AM
> > To: Quanah Gibson-Mount
> > Cc: openldap-software@OpenLDAP.org
> > Subject: Re: [openldap 2.1.12 + BerkeleyDb 4.1.24] db_checkpoint pb
> >
> > So it's sure if I choose to set the checkpointing directive in my
> > slapd.conf
> > like that: "checkpoint 250 15" rather than to use db_checkpoint in a
> > script
> > executed every hour by crontab ?? (cf. slapd.conf + DB_CONFIG)
> > Please, I would like to be sure because there is a FAQ which advises the
> > use of
> > the script rather than the checkpoint directive in the slapd.conf ?
>
> I don't know which FAQ you are referring to, so I can't comment on that. I
> can tell you, though, that this is the checkpointing/archiving method we
> recommend to our customers who are using our commercial builds, and they are
> quite happy with it.
>
> Your checkpoint statement will cause a checkpoint every 15 minutes or 250K
> of log data. This is four times more frequent than your crontab, so it may
> have an adverse impact on performance. OTOH, it provides a bit better
> recovery capability if your system should ever go down. The values you come
> up with for checkpoint need to be those that work best for the way you are
> using OpenLDAP.

Indeed, I noted that with a checkpoint every 15 minutes or 250K of log data (my
reference for these values is
"http://www.openldap.org/lists/openldap-software/200304/msg00683.html"; ??? ), my
mass-provisioning takes more time.
I think it's due to the parameter of my checkpoint directive  ... because I only
changed this point.
I tried with an other checkpoint: every 60 minutes or 1000K of log data too
but always the same pb: for creation of 5000 subscriptions it takes 2 hours
instead of few minutes ??
How to size my checkpoint to retrieve a correct performances ?
I'm going to try with a checkpoint every 60 minutes or 10000K of log data (max
size of my database log files = 10Mb) ...So I hope it will be ok

> In the end, all opinions are worth what you paid for them and Your Mileage
> May Vary ;-)
>
> Cheers,
>