[CentOS] Why slapd dying?

Fri Jan 19 02:47:34 UTC 2007
Jun Salen <nokijun at yahoo.com>

Thanks Johnny. My Slapd are now running after start
thru service start command. One of the reason why the
daemon is dying before is that the dbd database was
corrupted probably by sudden power off due to lose
power socket connection. I follow your advise and now
enable logging thru it. Thanks again and more power to
you and to Matt.

-------------------------

On Wed, 2007-01-17 at 20:02 -0500, Matt Hyclak wrote:
> On Thu, Jan 18, 2007 at 12:46:34AM +0000, Jun Salen
enlightened us:
> > I am wondering why slapd in my CentOS 4.4s erver
was
> > unable to run.  I already configure ldap to start
at
> > boot and when I issue command /sbin/service ldap
start
> > it is sucessfully started but again when I check
the
> > status etheir thru service or by netstat, it was
> > stopped and not exist respectively. Is anybody
> > encountered this. If you need some more info
> > just let me know. Thanks.
> > 
> 
> I ran into this after restoring the ldap database
files from a backup
> (/var/lib/ldap). You can create /etc/sysconfig/ldap
and in it put 
something
> like:
> 
> SLAPD_OPTIONS="-d XXX" 
> 
> and restart ldap. Man the slapd manpage for all the
options. If it is 
a
> corrupted database, you might look at the various
db_* commands, such 
as
> db_recover. Googling for any error messages you get
will help, too.

You can also turn on logging for slapd to figure out
want is going
on ...

1.  add this line to /etc/syslog.conf

local4.*                         /var/log/ldap.log

2. add this line to /etc/openldap/slapd.conf

loglevel 256

(there are numerous levels ... see the below link and
search the page
for loglevel)

http://www.openldap.org/doc/admin22/slapdconfig.html

set the loglevel back to a valid value (I use 0) when
finished debuging
based on the above link.

3. add this to /etc/logrotate.d/syslog

(somewhere in the log names line for syslog  {that is
the first line},
put this)

/var/log/ldap.log

(that will make ldap.log one of the logs it rotates)

----------------------
Some notes:

1.  The openldap people recommend a bdb type (and not
ldbm type)
database for the backend.  

Backing up the database with slapcat > filename   ... 
 and after 
making
sure that "filename" is OK, removing all the files in
/var/lib/ldap/ 
and
using slapadd -l filename to restore can fix database
issues.

you can also use slapcat > filename ... edit
slapd.conf to change from
ldbm to bdb database type ... create a DB_CONFIG file
in /var/lib/ldap/
and then do slapadd -l filename

you need to chown all files to ldap.ldap in
/var/log/ldap/ prior to
restarting ldap.

2.  look at the man pages for slapd_db_recover and
slapindex and use
those if you database is not good.

3.  setup a test machine and play with slapcat and
slapadd to get the
hang of it first with the slapcat output file.

4. Here is my DB_CONFIG and changes specifically to
slapd.conf for bdb
(if you are not using it now):

----------------
###DB_CONFIG###

#
# Set the database in memory cache size.
#
set_cachesize 0 52428800 0

# Automatically remove log files that are no longer
needed.
set_flags DB_LOG_AUTOREMOVE

#
# Set database flags.
# (for database loading/reindexing)
#set_flags       DB_TXN_NOSYNC
#set_flags DB_TXN_NOT_DURABLE

# Set log values.
#
set_lg_regionmax        1048576
set_lg_max              10485760
set_lg_bsize            2097152

------------------------
#slapd.conf adds#

#database ldbm
database bdb
cachesize 100000
checkpoint 512 720






junji
linux registered user #253162

Send instant messages to your online friends http://uk.messenger.yahoo.com