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

Problem with indexes not consistent



Hi,

I have a bit of problem keeping the indexes up to date on my LDAP
servers.

The LDAP servers are backends for a large mail service, that depends on
the maildrop field for each user. Often a single user cannot recieve
mail because that user cannot be lookup in the directory. This example
search shows the result from our three LDAP servers:

=======================================
# ./find 'prospero@*'
LDAP1:
LDAP2:
dn: cn=prospero@ofir.dk,ou=Members,o=Ofir
cn: prospero@ofir.dk

LDAP3:
dn: cn=prospero@ofir.dk,ou=Members,o=Ofir
cn: prospero@ofir.dk
=======================================

And an other example where a user is missing from a slave server:

=======================================
# ./find 'prospero2@*'
LDAP1:
dn: cn=prospero2@ofir.dk,ou=Members,o=Ofir
cn: prospero2@ofir.dk

LDAP2:
dn: cn=prospero2@ofir.dk,ou=Members,o=Ofir
cn: prospero2@ofir.dk

LDAP3:
=======================================

There is no error from the failing servers, the entry is just not found.

If I do a slapindex on the failing server, it finds all entries and
everuthing works again. So it seems like a indexing error. How do I make
it behave like its supposed to, and not corrupt the indexes? 

Has it something to do with the filesystem used? I had to use ldbm,
because of an error with bdb on reiserfs. I would like to try other
filesystems, but I can't really just try things out in this large live
setup...

	Casper Angelo
	OFiR, denmark



----------------
Our setup is like this:

LDAP1 (master)
	OpenLDAP: v.2.2.23
	Database: ldbm
	FS: reiserfs
	dist: Debian testing

LDAP2 & LDAP3 (slaves)
	OpenLDAP: v.2.2.23
	Database: ldbm
	FS: ext3
	dist: Debian testing

	Slaves are replicated from LDAP1