Re: pg_Restore

On Jan 21, 2013 7:17 AM, "bhanu udaya" <udayabhanu1984(at)hotmail(dot)com> wrote:
>
>
> Hello,
> Greetings !
> Thank you for the prompt reply. I have changed the settings as listed
below:
> > > shared_buffers = 1024MB
> > > work_mem = 512MB
> > > maintenance_work_mem = 512MB
> > > wal_buffers = 100MB
> > fsync = off # ONLY DURING INITIAL DATA LOAD!
> > checkpoint_segments = 128 # large value, such as 128 or 256 (16MB per
file, check disk space)
> > checkpoint_timeout = 30min
> > checkpoint_completion_target = 0.9
> > wal_level = minimal # You'll need to do a full base backup if you use
this
> But, have same problem. It is almost 1 hour now, the restoration is still
going on. After every test case execution, we would like to refresh the
database and expected refresh should be completed in less than 10 minutes.
Is this achievable with the kind of configuration I have listed in my
earlier email.
>
> Kindly help , as how to speed up this restoration process.
>

Try running pg_restore with the -1 option. If that doesn't help, try -m4 or
something like that (you'll have to remove the first option then, can't use
both at once)

But it's going to be pushing it anyway. Your scenario is going to create
thousands of files (assuming you have multiple tables in each of your
schemas as is normal), and that's just not something ntfs does very fast.
Once the files are there, I bet loading the data is reasonably fast since
it can't be all that big....

/Magnus