Hi

So just to go over what i have


server A (this is the original pgsql server 9.2)

Server X and Server Y ... PGSQL 9.6 in a cluster - streaming replication
with hot standby.


I have 2 tables about 2.5T of diskspace.

I want to get the date from A into X and X will replicate into Y.


I am currently on X using this command

pg_dump -U <USER> -h <Server A > -t BIGTABLE -a <DB> | sudo -u postgres -i
psql -q <DB>;

This is taking a long time, its been 2 days and I have xfered around 2T..
This is just a test to see how long and to populate my new UAT env. so I
will have to do it again.

Problem is time.  the pg_dump process is single threaded.
I have 2 routers in between A and X but its 10G networking - but my network
graphs don't show much traffic.

Server X is still in use, there are still records being inserted into the
tables.

How can I make this faster.

I could shutdown server A and present the disks to server X, could I load
this up in PGSQL and do a table to table copy - i presume this would be
faster ... is this possible ?  how do I get around the same DB name ?
What other solutions do I have ?

Alex




On 1 August 2017 at 23:24, Scott Marlowe <scott.marl...@gmail.com> wrote:

> On Mon, Jul 31, 2017 at 11:16 PM, Alex Samad <a...@samad.com.au> wrote:
> > Hi
> >
> > I double checked and there is data going over, thought I would correct
> that.
> >
> > But it seems to be very slow.   Having said that how do I / what tools
> do I
> > use to check through put
>
> Try the pg_current_xlog_location function on the slave?
>

Reply via email to