Stefan! This is like holding all your eyelids apart and shouting out
for somebody with a sharp stick! ;)
Watcha gonna do when you add a third server?
er.. lemme see... Multiply all OIDs by three and allocate new OIDs
modulo three?
Eeek!
Why not make the OID a composite of server ID and sequence? Even if
these were both packed in the same 64 bits, just so you keep it a
*little bit nasty*, and then maybe just get one eye poked out! ;)
(You could get away with a byte for the server ID, leaving masses of
bits for the item ID.)
Best wishes!
--r
On 22 May 2008, at 14:49, Stefan H. Holek wrote:
Hi Russ,
This is correct. I should have mentioned that the two servers are
configured so that server A creates odd, server B even sequence
numbers. This makes the duplicate id error pretty mysterious, IMO ;-)
Stefan
On 22.05.2008, at 13:15, Russ Ferriday wrote:
I have not read the reference, but it seems that this scheme can
never work without keys (OIDs) being qualified by which master they
originated on. Otherwise one of the mySQLs would need to be
responsible for handing out unique oids. You need an OID
origination namespace for each master.
--
Stefan H. Holek
[EMAIL PROTECTED]
Russ Ferriday - Topia Systems - Open Source content management with
Plone and Zope
[EMAIL PROTECTED] - office: +44 2076 1777588 - mobile: +44 7789 338868 -
skype: ferriday
_______________________________________________
For more information about ZODB, see the ZODB Wiki:
http://www.zope.org/Wikis/ZODB/
ZODB-Dev mailing list - ZODB-Dev@zope.org
http://mail.zope.org/mailman/listinfo/zodb-dev
