[Date Prev][Date Next] [Chronological] [Thread] [Top]

Re: Rif: Re: Slapd stops working after a NOT graceful shutdown

There have been discussion on the -devel list in the recent past about possible automated solutions to this problem.
The interesting thread is titled

back-bdb DB_RECOVER and soft restart

There, are outlined the pros and cons of automated recovery.
Many on this list can express their preference toward an administrator controlled recovery when using back-bdb. (I'm one of them)
back-ldbm is somewhat easier to mantain and so appear to be more reliable but I can assure you that even ldbm can become corrupt and then your only option is to slapadd.
At least with back-bdb you can try to recover from the transaction logs without loosing much if you checkpoint frequently enough.

Just my 2 cents

Walter Vendraminetto wrote:

I want to avoid DB failure after a crash: it looks like something
One of the replica servers uses OpenLDAP 2.0.27 with an earlier version of
BDB, and it has never had this kind of problem in the past.

I'm wondering if there is a configuration issue to take into account in the
BDB 4.1.25 in order to prevent such situation.


-------Messaggio originale-------

Da: Medievalist
Data: lunedì 09 febbraio 2004 19.19.43
A: openldap-software@OpenLDAP.org
Oggetto: Re: Slapd stops working after a NOT graceful shutdown

Personally, I'm *glad* OpenLDAP doesn't want to work with a corrupted

Here are two possible solutions to your problem:

1) Every time you modify your LDAP database, make a fresh backup, and build a new fresh database from the most recent backup every time you boot up.
This will increase your bootup time and won't address any other problems.

2) Buy an uninteruptible power supply. This will solve other problems you haven't noticed yet. *nix software is not designed to survive powerfails.

Note, your hardware will eventually burn up from voltage spikes caused by
power failures if you don't get some kind of power conditioning. So if you think
an UPS is too expensive, you are probably fooling yourself!

If you have more questions about recovering corrupted databases you need to
ask the vendors of your database (Red Hat or more likely Sleepycat). Your
OpenLDAP software appears to be working correctly.


On 9 Feb 2004 at 18:24, Walter Vendraminetto wrote:

Date sent: Mon, 9 Feb 2004 18:24:02 +0100 (ora solare Europa occidentale)
From: "Walter Vendraminetto" <lists@zonatex.net>
To: "OpenLDAP List" <openldap-software@OpenLDAP.org>
Subject: Slapd stops working after a NOT graceful shutdown
Priority: non-urgent

I'd like to solve a problem which I faced few months ago.

When the server (RH 7.3) is shut down because of power black-out (UPS not
working properly) or brutal reset the LDAP service does not work anymore.


have to slapcat the database and rebuild it to make the service work again

Even if it does not happen often (last time this morning) I want to fix it

I think it's a matter concerning the back-end configuration but I'm in
trouble on how to proceed: the only option set in the DB_CONFIG is the set_cachesize'.
How can I debug it ? I saved the 'corrupted' DB files ...

thank you very much

Walter Vendraminetto - Research Consultant
Centro di Competenza per la Progettazione di Sistemi Dedicati
c/o Dipartimento di Informatica - Universita` di Verona
Strada le Grazie, 15 - 37134 Verona (Italy)
email: vendra@sci.univr.it
tel: 045-802-7062
fax: 045-802-7068

SysNet - via Dossi,8 27100 Pavia Tel: +390382573859 Fax: +390382476497