Bruce Momjian napsal(a):
> Heikki Linnakangas wrote:
>> Bruce Momjian wrote:
>>> As far as the page not fitting after conversion, what about some user
>>> command that will convert an entire table to the new format if page
>>> expansion fails.
>> VACUUM?
>>
>> Having to run a manual command defeats the purpose somewhat, though.
>> Especially if you have no way of knowing on what tables it needs to be
>> run on.
>
> My assumption is that the page not fitting would be a rare case so
> requiring something like vacuum to fix it would be OK.
It is 1-2% records per heap. I assume that is is more for BTree.
> What I don't want to do it to add lots of complexity to the code just to
> handle the page expansion case, when such a case is rare and perhaps can
> be fixed by a vacuum.
Unfortunately it is not so rare. And only heap on 32bit x86 platform (4byte Max
alignment) is no problem. But all index pages are affected.
>
>> In fact, I don't think there's any low-level data format changes yet
>> between 8.3 and 8.4, so this would be a comparatively easy release to
>> implement upgrade-in-place. There's just the catalog changes, but AFAICS
>> nothing that would require scanning through relations.
>
> Yep.
I did not test now but pg_upgrade.sh script worked fine in May without any
modification for conversion 8.3->8.4.
Zdenek
--
Zdenek Kotala Sun Microsystems
Prague, Czech Republic http://sun.com/postgresql
