On Mon, Oct 3, 2011 at 8:21 AM, Heikki Linnakangas
<heikki.linnakangas@enterprisedb.com> wrote:
> On 29.09.2011 14:31, Fujii Masao wrote:
>>
>> On Tue, Sep 27, 2011 at 8:06 PM, Florian Pflug<fgp@phlo.org> wrote:
>>>
>>> Actually, why don't we use that machinery to implement this? There's
>>> currently no rm_safe_restartpoint callback for RM_XLOG_ID, so we'd just need
>>> to create one that checks whether invalid_page_tab is empty.
>>
>> Okay, the attached patch prevents the creation of restartpoints by using
>> rm_safe_restartpoint callback if we've not reached a consistent state yet
>> and the invalid-page table is not empty. But the invalid-page table is not
>> tied to the specific resource manager, so using rm_safe_restartpoint for
>> that seems to slightly odd. Is this OK?
>
> I don't think this should use the rm_safe_restartpoint machinery. As you
> said, it's not tied to any specific resource manager. And I've actually been
> thinking that we will get rid of rm_safe_restartpoint altogether in the
> future. The two things that still use it are the b-tree and gin, and I'd
> like to change both of those to not require any post-recovery cleanup step
> to finish multi-page operations, similar to what I did with GiST in 9.1.
I thought that was quite neat doing it that way, but there's no
specific reason to do it that way I guess. If you're happy to rewrite
the patch then I guess we're OK.
I certainly would like to get rid of rm_safe_restartpoint in the
longer term, hopefully sooner.
--
Simon Riggs http://www.2ndQuadrant.com/
PostgreSQL Development, 24x7 Support, Training & Services