Fedora directory users - May 2009 - HELP! Lost single Master

Hello, everyone.  I did something really stupid.  I accidentally
destroyed our Single Master replica server (that''s what I get for
resynching a failed RAID1 drive without realizing the drive device names
had changed because of the failed drive!).  Thankfully we were not yet
in production (although very close and thus using lots of data).
However, because we were not yet in production, it was not backed up.

I''d like to recover the data from the read only replica.  I''m
guessing I
can disable replication on the RO server and it will retain the data but
now with write privileges.  I can then rebuild the master, create a
reverse replication agreement to transfer the data back.  Then I can
break that replication agreement, create a new one in the correct
direction, and reinitialize the supplier.

However, when I created the replica server, we told it to join an
existing directory server domain.  Is this simply based upon name? In
other words, once I rebuild the main server and configure it using the
same name and address as the original, will the secondary recognize it
as the domain to which it registered or do I really need to destroy all
servers and recreate the entire domain from scratch? Thanks - John
-- 
John A. Sullivan III
Open Source Development Corporation
+1 207-985-7880
jsullivan@opensourcedevel.com

http://www.spiritualoutreach.com
Making Christianity intelligible to secular society

John A. Sullivan III wrote:
> Hello, everyone.  I did something really stupid.  I accidentally
> destroyed our Single Master replica server (that''s what I get for
> resynching a failed RAID1 drive without realizing the drive device names
> had changed because of the failed drive!).  Thankfully we were not yet
> in production (although very close and thus using lots of data).
> However, because we were not yet in production, it was not backed up.
>
> I''d like to recover the data from the read only replica. 
I''m guessing I
> can disable replication on the RO server and it will retain the data but
> now with write privileges.  I can then rebuild the master, create a
> reverse replication agreement to transfer the data back.  Then I can
> break that replication agreement, create a new one in the correct
> direction, and reinitialize the supplier.
>   
No, it''s not that complicated.  The simplest way would be to just use 
db2ldif to dump your replica database, then use ldif2db to import it 
into the master.  Then reinit your replica from the
master.
> However, when I created the replica server, we told it to join an
> existing directory server domain.  Is this simply based upon name? In
> other words, once I rebuild the main server and configure it using the
> same name and address as the original, will the secondary recognize it
> as the domain to which it registered or do I really need to destroy all
> servers and recreate the entire domain from scratch? Thanks - John
>   
I''m not sure what you mean.

On Mon, 2009-05-04 at 08:28 -0600, Rich Megginson wrote:
> John A. Sullivan III wrote:
> > Hello, everyone.  I did something really stupid.  I accidentally
> > destroyed our Single Master replica server (that''s what I get
for
> > resynching a failed RAID1 drive without realizing the drive device
names
> > had changed because of the failed drive!).  Thankfully we were not yet
> > in production (although very close and thus using lots of data).
> > However, because we were not yet in production, it was not backed up.
> >
> > I''d like to recover the data from the read only replica. 
I''m guessing I
> > can disable replication on the RO server and it will retain the data
but
> > now with write privileges.  I can then rebuild the master, create a
> > reverse replication agreement to transfer the data back.  Then I can
> > break that replication agreement, create a new one in the correct
> > direction, and reinitialize the supplier.
> >   
> No, it''s not that complicated.  The simplest way would be to just
use
> db2ldif to dump your replica database, then use ldif2db to import it 
> into the master.  Then reinit your replica from the master.
Phew!
> > However, when I created the replica server, we told it to join an
> > existing directory server domain.  Is this simply based upon name? In
> > other words, once I rebuild the main server and configure it using the
> > same name and address as the original, will the secondary recognize it
> > as the domain to which it registered or do I really need to destroy
all
> > servers and recreate the entire domain from scratch? Thanks - John
> >   
> I''m not sure what you mean.
When we installed the first server, we told it there was no existing
directory server instance with which to register.  When we created the
RO replica server, we told it to register with the existing instance.

How do we handle that now that the original instance has been destroyed?
What did registering it do? Thanks - John
<snip>
-- 
John A. Sullivan III
Open Source Development Corporation
+1 207-985-7880
jsullivan@opensourcedevel.com

http://www.spiritualoutreach.com
Making Christianity intelligible to secular society

John A. Sullivan III wrote:
> On Mon, 2009-05-04 at 08:28 -0600, Rich Megginson wrote:
>   
>> John A. Sullivan III wrote:
>>     
>>> Hello, everyone.  I did something really stupid.  I accidentally
>>> destroyed our Single Master replica server (that''s what I
get for
>>> resynching a failed RAID1 drive without realizing the drive device
names
>>> had changed because of the failed drive!).  Thankfully we were not
yet
>>> in production (although very close and thus using lots of data).
>>> However, because we were not yet in production, it was not backed
up.
>>>
>>> I''d like to recover the data from the read only replica. 
I''m guessing I
>>> can disable replication on the RO server and it will retain the
data but
>>> now with write privileges.  I can then rebuild the master, create a
>>> reverse replication agreement to transfer the data back.  Then I
can
>>> break that replication agreement, create a new one in the correct
>>> direction, and reinitialize the supplier.
>>>   
>>>       
>> No, it''s not that complicated.  The simplest way would be to
just use
>> db2ldif to dump your replica database, then use ldif2db to import it 
>> into the master.  Then reinit your replica from the master.
>>     
> Phew!
>   
>>> However, when I created the replica server, we told it to join an
>>> existing directory server domain.  Is this simply based upon name?
In
>>> other words, once I rebuild the main server and configure it using
the
>>> same name and address as the original, will the secondary recognize
it
>>> as the domain to which it registered or do I really need to destroy
all
>>> servers and recreate the entire domain from scratch? Thanks - John
>>>   
>>>       
>> I''m not sure what you mean.
>>     
> When we installed the first server, we told it there was no existing
> directory server instance with which to register.  When we created the
> RO replica server, we told it to register with the existing instance.
>
> How do we handle that now that the original instance has been destroyed?
> What did registering it do? Thanks - John
>   
Ah, ok.  I think you need to run the register-ds-admin.pl command after 
you restore the data on the master, and tell it you want the master to 
be the configuration directory server.

Then you''ll have to run setup-ds-admin.pl -u (or possibly 
register-ds-admin.pl) on the replica to register it with the 
configuration directory server.
> <snip>
>