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.
> However, when I created the replica server, we told it to join
> 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
John A. Sullivan III
Open Source Development Corporation
Making Christianity intelligible to secular society