Re: Parallel Bitmap Heap Scan reports per-worker stats in EXPLAIN ANALYZE

On Fri, Jan 20, 2023 at 2:04 PM David Geier <geidav(dot)pg(at)gmail(dot)com> wrote:
>
> Hi hackers,
>
> EXPLAIN ANALYZE for parallel Bitmap Heap Scans currently only reports
> the number of heap blocks processed by the leader. It's missing the
> per-worker stats. The attached patch adds that functionality in the
> spirit of e.g. Sort or Memoize. Here is a simple test case and the
> EXPLAIN ANALYZE output with and without the patch:
>

> With the patch:
>
> Gather (actual rows=99501 loops=1)
> Workers Planned: 2
> Workers Launched: 2
> -> Parallel Bitmap Heap Scan on foo (actual rows=33167 loops=3)
> Recheck Cond: ((col0 > 900) OR (col1 = 1))
> Heap Blocks: exact=98
> Worker 0: Heap Blocks: exact=171 lossy=0
> Worker 1: Heap Blocks: exact=172 lossy=0

else
{
+ if (planstate->stats.exact_pages > 0)
+ appendStringInfo(es->str, " exact=%ld", planstate->stats.exact_pages);
+ if (planstate->stats.lossy_pages > 0)
+ appendStringInfo(es->str, " lossy=%ld", planstate->stats.lossy_pages);
appendStringInfoChar(es->str, '\n');
}
}
....
+ for (int n = 0; n < planstate->shared_info->num_workers; n++)
+ {
....
+ "Heap Blocks: exact="UINT64_FORMAT" lossy=" INT64_FORMAT"\n", +
si->exact_pages, si->lossy_pages);

Shouldn't we use the same format for reporting exact and lossy pages
for the actual backend and the worker? I mean here for the backend you
are showing lossy pages only if it is > 0 whereas for workers we are
showing 0 lossy pages as well.

--
Regards,
Dilip Kumar
EnterpriseDB: http://www.enterprisedb.com