Re: pg_restore speed

"Joshua D. Drake" <jd(at)commandprompt(dot)com> writes:
> Colton A Smith wrote:
>> I have a 90G postgis-enabled postgresql
>> database that I'm upgrading from 8.1.5 to
>> 8.2.1. I initiated pg_restore nearly
>> six days ago and it's still churning away,
>> surely but slowly. Is this amount of
>> time normal? Or have I done something
>> wrong?

> 6 days seems like an awful long time I could see 12-15 hours. What type
> of machine do you have? What was the pg_dump command you used and
> respective pg_restore?

What is the restore actually doing at the moment? (Look in
pg_stat_activity)

For such a large DB I'd think you'd really need to increase
shared_buffers, checkpoint_segments, and maintenance_work_mem well
beyond their default values ... did you do that before starting to
restore? (Actually I think you could probably do the latter two
on-the-fly without disturbing the restore, so that might be something
to try before giving up and starting over.)

regards, tom lane