On Wed, 3 Apr 2019 08:33:54 -0500
Ron <ronljohnsonjr@gmail.com> wrote:
> On 4/3/19 8:18 AM, Steven Lembark wrote:
> > Trying to find a way of moving a large table between databases
> > in the same cluster. There is not sufficient space to copy the
> > contents -- the dedicated tablespace that fits the beastie is
> > on an 80% full disk.
> >
> > Given that the two databases live in the same cluster and have
> > the owner & the tablespace in common, is there any way to move
> > the contents without a dump & reload?
>
> COPY TO STDIN and pipe that through ssh to another server?
That makes a copy. There is no place to store two copies of the
full data on disk.
I'm basically trying to perform the moral equivalent of what "mv"
does in the filesystem: re-link the inode and remove the old inode
without duplicating the data.
Since the tablespace is visible to both I would get that there is
some way to alter table... that allocates the space on disk to the
new database w/o having to unload all 400M rows -- if it comes down
to that I'll have to copy it into xz and load it back in the new
database.
--
Steven Lembark 3920 10th Ave South
Workhorse Computing Birmingham, AL 35222
lembark@wrkhors.com +1 888 359 3508