Re: pgsql: Skip full index scan during cleanup of B-tree indexes when possi

On Wed, Apr 18, 2018 at 9:21 PM, Peter Geoghegan <pg(at)bowt(dot)ie> wrote:

> On Wed, Apr 18, 2018 at 11:12 AM, Fujii Masao <masao(dot)fujii(at)gmail(dot)com>
> wrote:
> > On Thu, Apr 5, 2018 at 1:29 AM, Teodor Sigaev <teodor(at)sigaev(dot)ru> wrote:
> >> Skip full index scan during cleanup of B-tree indexes when possible
> >
> > This commit added XLOG_BTREE_META_CLEANUP, so btree_desc() and
> > btree_identify() should be updated so that they handle
> XLOG_BTREE_META_CLEANUP.
> > But ISTM that you forgot doing that.
>

Thank you for notice. See attached
bt-vacuum-cleanup-wal-record-desc-identify.patch fixing that.

Another thing that I noticed is that the metapage stores
> btm_last_cleanup_num_heap_tuples as a float4, even though
> xl_btree_metadata stores it as a double. I suggest that both places
> store it as float8, to be consistent. (It should not be double because
> we always avoid using anything other types with explicit typedef'd
> widths in WAL records.)

Good catch, thank you! I also agree that both these fields should be of
float8 type.
Please, find attached bt-vacuum-cleanup-float8-num-heap-tuples.patch fixing
that.

------
Alexander Korotkov
Postgres Professional: http://www.postgrespro.com
The Russian Postgres Company