> If Oracle announced that they were releasing a patch for six critical
> data-loss bugs in 9i, one of which affected you, would you call support
> and ask them to make a custom patch just for you that only fixed the one
> you had encountered?
Oh, and when they say they can't guarantee that fix will work without
the others, and won't do any QA on it, and that it'll have to be built
using a non-standard and untested compilation environment instead of
their regular update builder, would you still want to go ahead? Even
though, by trying to extract just that one patch, you're actually
changing the whole program completely, making a much bigger, much less
well tested change than you would be by applying the whole official patch?
To me, it makes zero sense.
I think Tom Lane said it best here:
http://www.mail-archive.com/pgsql-general@postgresql.org/msg93045.html
"The real bottom line here, and one I'll reiterate every chance I get,
is that we don't make updates to back branches because we're too bored
to have anything else to do. If you're on 8.1.5, and the current
release in that branch is 8.1.8, then you're missing some bug fixes
that are probably significant."
Anyway, if you want to suffer through doing this yourself, I *think* the
change in question is from 8.1.7, documented in the release notes as:
"Fix autovacuum to avoid leaving non-permanent transaction IDs in
non-connectable databases (Alvaro). This bug affects the 8.1 branch only."
http://www.postgresql.org/docs/8.1/static/release-8-1-7.html
That should give you something to search the cvs logs for, so you can
extract the patch and apply it to your version.
Here's some documentation on how to get the source code:
http://www.postgresql.org/docs/8.1/static/cvs.html
You can use cvs commands, or a graphical cvs program, to browse the
change history in order to locate the specific change you need, extract
it as a diff, and apply it to a checkout of 8.1.2 cvs.
Some general information on PostgreSQL development:
http://www.postgresql.org/developer/coding
Some related discussion:
http://www.mail-archive.com/pgsql-general@postgresql.org/msg93045.html
http://archives.postgresql.org/pgsql-hackers/2006-03/msg01294.php
http://grokbase.com/topic/2007/08/22/general-could-not-open-file-pg-clog-0bff/sqm2TnwMqn3Aqy-ZXu9e83Ve3jM
http://www.mail-archive.com/pgsql-general@postgresql.org/msg93022.html
.... and I've wasted half an hour on this for no good reason.
--
Craig Ringer