Re: pgbench: Skipping the creating primary keys after initialization

On Thu, Aug 3, 2017 at 11:31 PM, Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us> wrote:
> Masahiko Sawada <sawada(dot)mshk(at)gmail(dot)com> writes:
>> If we want to create other tables and load data to them as we want we
>> can do that using psql -f. So an alternative ways is having a flexible
>> style option for example --custom-initialize = { [load, create_pkey,
>> create_fkey, vacuum], ... }. That would solve this in a better way.
>
> FWIW, I like that significantly better than your original proposal.
> It'd allow people to execute parts of pgbench's standard initialization
> sequence and then do other things in between (in psql). Realistically,
> that's probably about as much win as we need here --- if you're veering
> far enough away from the standard scenario that that doesn't do it for
> you, you might as well just write an all-custom setup script in psql.
>

Attached patch introduces --custom-initialize option that allows us to
specify the initialization step and its order. For example, If you
want to skip building primary keys you can specify
--custom-initialize="create_table, load_data, vacuum". Since each
custom initialization commands is invoked in specified order, for
example we also can create primary keys *before* loading data. The
data-generation is doing on client side, so progress information for
initialization is still supported.

Regards,

--
Masahiko Sawada
NIPPON TELEGRAPH AND TELEPHONE CORPORATION
NTT Open Source Software Center