Re: 16-bit page checksums for 9.2

On 29.02.2012 21:18, Alvaro Herrera wrote:
> Excerpts from Heikki Linnakangas's message of mié feb 29 16:09:02 -0300 2012:
>> I thought my view on how this should be done was already clear, but just
>> in case it isn't, let me restate: Enlarge the page header to make room
>> for the checksum. To handle upgrades, put code in the backend to change
>> the page format from old version to new one on-the-fly, as pages are
>> read in. Because we're making the header larger, we need to ensure that
>> there's room on every page. To do that, write a utility that you run on
>> the cluster before running pg_upgrade, which moves tuples to ensure
>> that. To ensure that the space doesn't get used again before upgrading,
>> change the old version so that it reserves those N bytes in all new
>> insertions and updates (I believe that approach has been discussed
>> before and everyone is comfortable with backpatching such a change). All
>> of this in 9.3.
>
> Note that if we want such an utility to walk and transform pages, we
> probably need a marker in the catalogs somewhere so that pg_upgrade can
> make sure that it was done in all candidate tables -- which is something
> that we should get in 9.2 so that it can be used in 9.3.

In the simplest form, the utility could just create a magic file in the
data directory to indicate that it has run. All we need is a boolean
flag, unless you want to be fancy and make the utility restartable.
Implemented that way, there's no need to have anything in the catalogs.

> Such a marker would also allow us get rid of HEAP_MOVED_IN and
> HEAP_MOVED_OUT.

True.

--
Heikki Linnakangas
EnterpriseDB http://www.enterprisedb.com