On Wed, Dec 18, 2013 at 11:35 AM,  <[email protected]> wrote:
> I've got PG 9.3 and wal-e (6.6) doing replication on a pretty basic setup (1
> master, 1 slave, 1 read replica).
>
> If master goes down and I switch to slave, then I also update the read
> replica to start talking to the slave. That all works fine. However, what
> are some best practices at this point? For example:
>
> 1. Should I do an immediate base backup (push) of the slave since my logs
> are now reset?

Yes.  I plan to allow tying together two such prefixes in a future
release, but I suggest this is the best option for now.

> 2. Should that push live in a new directory on S3 or can it live in the same
> spot I was pushing to with master?

New place, otherwise you risk overwrite.

> 3. Should I also change where wal files live?

You should use a new prefix for all of the above, otherwise you risk
confusing and dangerous WAL overwriting.

> 4. What should I prune?

After you are satisfied with the failover for a little while, you can
delete the old prefix in its entirety.

Does that all seem reasonable?

-- 
You received this message because you are subscribed to the Google Groups 
"wal-e" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to [email protected].
For more options, visit https://groups.google.com/groups/opt_out.

Reply via email to