Thread: Variable Block Size Dilemma

Variable Block Size Dilemma

From
darrenk@insightdist.com (Darren King)
Date:
I'm stuck as to how to make this a parameter for the backend.

The postmaster could read a pg_blocksize file and then pass that,
but how could _that_ be set differently when calling the same
copy of initdb?

I'd like to avoid having to set it in the Makefile in the way
that the NAMEDATALEN and OIDNAMELEN variables are set up.

*small lightbulb on*

Would it be acceptable for this to be in a file in the PGLIB
directory?  This could be read in by initdb and then passed along
to the postgres calls.  Does this seem reasonable and/or clean?

Or is there a really obvious way that I'm just not seeing right now?

darrenk

Re: [HACKERS] Variable Block Size Dilemma

From
Bruce Momjian
Date:
>
>
> I'm stuck as to how to make this a parameter for the backend.
>
> The postmaster could read a pg_blocksize file and then pass that,
> but how could _that_ be set differently when calling the same
> copy of initdb?
>
> I'd like to avoid having to set it in the Makefile in the way
> that the NAMEDATALEN and OIDNAMELEN variables are set up.
>
> *small lightbulb on*
>
> Would it be acceptable for this to be in a file in the PGLIB
> directory?  This could be read in by initdb and then passed along
> to the postgres calls.  Does this seem reasonable and/or clean?
>
> Or is there a really obvious way that I'm just not seeing right now?

You ask about the block size as part of initdb, and create a file in
/pgsql/data as part of initdb's work.  Then each postmaster reads the
file, and passes the value to each backend as a parameter.

--
Bruce Momjian
maillist@candle.pha.pa.us