[389-users] glue entry problem

[Elizabeth Jones]

>
> You mentioned 2 servers ldap1 and ldap2.  Are they both masters? You put
> "local consumer" to ldap2.  Does that mean ldap2 is a read only replica?

We have two data centers and each data center has an ldap1 and ldap2.  All
4 are masters, but we only ever send updates to DCA-ldap1.  That then
pushes to DCA-ldap2 and to DCB-ldap1.  DCB-ldap1 pushes to DCB-ldap2, so
the ldap2's are really functioning as consumers all the time.

>
> It looks to me ldap1 got broken and ldap2 is still healthy.  You may
> want to make ldap1 in sync with ldap2 and start from there. If ldap2 is
> a master,  you could re-initialize ldap1 from ldap2.
>
> If ldap2 is a read only replica, you could export the contents with
> db2ldif -r -n <your_backend> command line utility on ldap2 and import
> the exported ldif file to ldap1.
>
> Or if you don't mind losing the replication information such as
> tombstones and state info, you could export the contents without "-r" by
> db2ldif on ldap2 and import the ldif file to ldap1, then re-initialize
> ldap2 on ldap1.
>
> Hope you could choose one of the 3 ways and it fixes the problem.

I will give this a shot.  When we tried to initialize these last week the
initialization emptied out the receiving ldap and I believe that was a bug
that is fixed in 1.3.  I recovered by using a backup but I don't want to
try initializing again until we can upgrade to 1.3. Can I still do the
ldif on ldap2, since it is actually a master rather than a read only
replica?

Thanks,
EJ