Hi folks,
please forgive what feels like a no-brainer even as I ask it, but...
I've read dozens of times in these lists that when one is upgrading from
an older to newer version of PG, the DB being dumped (older version)
should be done so using pg_dump from the newer version. I think I've
probably absorbed that bit of wisdom, but I've never actually had to put
it to use so it's always just been an academic understanding and a quick
scan every time I see the topic come up.
Well I'm actually going to need to dump an older DB and restore to a
newer version very soon, and I'd just like a point of clarification
please. Which bit exactly are we supposed to use from the newer version?
I mean, I've used pg_dump hundreds of times but I've never had need to
actually look at it and I've always assumed that it been a binary file,
especially since the advise is always to use the newer version's pg_dump
to start the migration, intimating the working bits are inside there
somewhere, and that they're different in every version.
Unless my installation is unique in some way of which I'm yet unaware,
pg_dump seems to be just a handful of lines in a perl script. In fact,
pg_dump, pg_restore and pg_dumpall are all simlinks to the same simple
perl script, the contents of which seem to be identical in both my
'older' and 'newer' versions of PG. Does this mean I can trust any old
dump from my older server to seamlessly plug into my newer version? Or
does thin mean there are other 'gizmos' than pg_dump which I need to
copy from new machine to old machine to perform the dump?
Andy
