Re: WIP: Avoid creation of the free space map for small tables

On Tue, Jan 29, 2019 at 5:59 AM Amit Kapila <amit(dot)kapila16(at)gmail(dot)com> wrote:
>
> On Tue, Jan 29, 2019 at 12:37 AM John Naylor
> <john(dot)naylor(at)2ndquadrant(dot)com> wrote:
> > > I think here you need to clear the map if it exists or clear it
> > > unconditionally, the earlier one would be better.
> >
> > Ok, maybe all callers should call it unconditonally, but within the
> > function, check "if (FSM_LOCAL_MAP_EXISTS)"?
> >
>
> Sounds sensible. I think we should try to reproduce these failures,
> for ex. for pgbench failure, we can try the same test with more
> clients.
>

I am able to reproduce this by changing pgbench test as below:

--- a/src/bin/pgbench/t/001_pgbench_with_server.pl
+++ b/src/bin/pgbench/t/001_pgbench_with_server.pl
@@ -56,9 +56,9 @@ $node->safe_psql('postgres',
'CREATE UNLOGGED TABLE insert_tbl (id serial primary key); ');

pgbench(
- '--no-vacuum --client=5 --protocol=prepared --transactions=25',
+ '--no-vacuum --client=10 --protocol=prepared --transactions=25',
0,
- [qr{processed: 125/125}],
+ [qr{processed: 250/250}],

You can find this change in attached patch. Then, I ran the make
check in src/bin/pgbench multiple times using test_conc_insert.sh.
You can vary the number of times the test should run, if you are not
able to reproduce it with this.

The attached patch (clear_local_map_if_exists_1.patch) atop the main
patch fixes the issue for me. Kindly verify the same.

--
With Regards,
Amit Kapila.
EnterpriseDB: http://www.enterprisedb.com