Re: forcing a rebuild of the visibility map

On Thu, Jun 16, 2016 at 10:03 PM, Robert Haas <robertmhaas(at)gmail(dot)com> wrote:
> On Thu, Jun 16, 2016 at 2:33 AM, Masahiko Sawada <sawada(dot)mshk(at)gmail(dot)com> wrote:
>> Option name DISABLE_PAGE_SKIPPING is good to me.
>> I'm still working on this, but here are some review comments.
>>
>> ---
>> +CREATE FUNCTION pg_truncate_visibility_map(regclass)
>> +RETURNS void
>> +AS 'MODULE_PATHNAME', 'pg_truncate_visibility_map'
>> +LANGUAGE C STRICT
>> +PARALLEL UNSAFE; -- let's not make this any more dangerous
>> +
>>
>> "REVOKE ALL ON FUNCTION pg_truncate_visibility_map(regclass) FROM
>> PUBLIC;" is missing in pg_visibility/pg_visibility--1.0--1.1.sql.
>
> OK, thanks. I'll fix that.
>
>> I think that VACUUM with VERBOSE option can show the information for
>> how many frozen pages we skipped like autovacuum does. Thought?
>> Patch attached.
>
> I'm not sure. The messages VACUUM emits are already quite long and
> hard to read, and adding more lines to them might make that problem
> worse. On the other hand, having more details can be helpful, too.

Because these informations are emitted only when VERBOSE option is
specified, I think that it would be better to output that rather than
nothing, although it makes messages more complicated.

Regards,

--
Masahiko Sawada