Re: Updating FSM on recovery

From: Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us>
To: Heikki Linnakangas <heikki(dot)linnakangas(at)enterprisedb(dot)com>
Cc: PostgreSQL-development <pgsql-hackers(at)postgresql(dot)org>
Subject: Re: Updating FSM on recovery
Date: 2008-10-29 12:36:54
Message-ID: 11303.1225283814@sss.pgh.pa.us
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-hackers

Heikki Linnakangas <heikki(dot)linnakangas(at)enterprisedb(dot)com> writes:
> The FSM would be perfectly happy to just initialize torn or otherwise
> damaged pages, so I think we should add yet another mode to ReadBuffer()
> to allow that. We could also treat read() errors as merely warnings in
> that mode, effectively the same as with zero_damaged_pages=on.

> The ReadBuffer() interface is already pretty complex, with all the
> different variants. We should probably keep the good old ReadBuffer()
> the same, for the sake of simplicity in the callers, but try to reduce
> the number of other variatns.

Indeed. Did you see the discussion about the similarly-too-complex
heap_insert API a couple days ago in connection with bulk-write
scenarios? The conclusion there was to try to shift stuff into a
bitmask options argument, in hopes that future additions might not
require touching every caller. Can we do it similarly here?

regards, tom lane

In response to

Responses

Browse pgsql-hackers by date

  From Date Subject
Next Message Tom Lane 2008-10-29 12:44:07 Re: some problem with casting unknown to smallint
Previous Message Pavel Stehule 2008-10-29 12:31:46 Re: some problem with casting unknown to smallint