Re: automatic analyze: readahead - add "IO read time" log message

Hi,

On 2/10/21 11:10 PM, Stephen Frost wrote:
> Greetings,
>
> * Heikki Linnakangas (hlinnaka(at)iki(dot)fi) wrote:
>> On 05/02/2021 23:22, Stephen Frost wrote:
>>> Unless there's anything else on this, I'll commit these sometime next
>>> week.
>>
>> One more thing: Instead of using 'effective_io_concurrency' GUC directly,
>> should call get_tablespace_maintenance_io_concurrency().
>
> Ah, yeah, of course.
>
> Updated patch attached.
>

A couple minor comments:

1) I think the patch should be split into two parts, one adding the
track_io_timing, one adding the prefetching.

2) I haven't tried but I'm pretty sure there'll be a compiler warning
about 'prefetch_maximum' being unused without USE_PREFETCH defined.

3) Is there a way to reduce the amount of #ifdef in acquire_sample_rows?

This makes the code rather hard to read, IMHO. It seems to me we can
move the code around a bit and merge some of the #ifdef blocks - see the
attached patch. Most of this is fairly trivial, with the exception of
moving PrefetchBuffer before table_scan_analyze_next_block - AFAIK this
does not materially change the behavior, but perhaps I'm wrong.

regards

--
Tomas Vondra
EnterpriseDB: http://www.enterprisedb.com
The Enterprise PostgreSQL Company