[Bucardo-general] bucardo 5 master-master sequence odd/even?

[Greg Sabino Mullane]

On Mon, Aug 25, 2014 at 09:05:14AM -0400, Jonathan Brinkman wrote:
> But I thought the B5 upgrade meant that multi-master scenarios were handled
> internally. What if I have a sync that includes (say) 5 masters? Would I
> need to alter the sequences accordingly to avoid this primary-key conflict
> problem (INCREMENT=5)? 

Yes, that would be one solution. Doesn't scale well of course, because you 
are stuck with X number of servers due to the increment.

> Also I had considered using a sequence strategy where each master's sequence
> started with a different number (1-9 say), but that doesn't work if Bucardo
> is checking/modifying the sequences for MAX on each restart. And would of
> course be limited to 9 masters.

Right. If you are using tricks to have different sequences on each server, 
you do NOT want Bucardo to track the sequences. A better sequence strategy 
than the inrement trick would be to reserve the first two numbers for a potential 
pool of 99 servers, or to simply use pre-defined ranges. As sequences can be BIGINT, 
there is little danger of running out of values. This will involve setting a 
different MINVALUE and MAXVALUE for each server.

-- 
Greg Sabino Mullane greg at endpoint.com
End Point Corporation
PGP Key: 0x14964AC8
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 163 bytes
Desc: Digital signature
URL: <https://mail.endcrypt.com/pipermail/bucardo-general/attachments/20140825/eafffc3a/attachment.sig>