Occassional index corrupts

["Vito Chan" <vito.chan@hk.so-net.com.hk>]

Dear all,

We have a openldap server (2.0.7) running on redhat linux 6.2. Our
servers have frequent read/write and we find that some records cannot be
searched exactly. (We have build an index uid eq)

e.g. we can search for uid=*abc while not uid=abc. The problem can be
solved by running slapcat but the problem will happen soon after the
ldap has a few more writes. This problem cause us so much trouble since
most of the user cannot search their account occassionaly and we cannot
always rebuild the index since it will cause so much service
interruption. 

I read from the mailing list that some of the problems may be caused by
gdbm backend (We are now using gdbm) and it will solved by upgrading the
openldap to 2.1.x and using bdb backend. But I am not sure whether it
really works since these problems occur randomly,

Can any expert here can share their experience on this problem and how
they can solve this ?

Best Regards,

Vito