Re: Sudden FTS-related error from parallel worker in 9.6

More observations. I've fixed that issue by obvious change -- commenting
out "default_text_search_config" in postgresql.conf. The error gone.

Then I altered database, placing incorrect FTS default configuration back:

test=# alter database test set default_text_search_config to 'ru';
NOTICE:  text search configuration "ru" does not exist
ALTER DATABASE


Now the query with aggregation produces only warnings (from each worker),
no errors:

test=#  explain analyze select min(ts) from test;
WARNING:  invalid value for parameter "default_text_search_config": "ru"
WARNING:  invalid value for parameter "default_text_search_config": "ru"
WARNING:  invalid value for parameter "default_text_search_config": "ru"
WARNING:  invalid value for parameter "default_text_search_config": "ru"
WARNING:  invalid value for parameter "default_text_search_config": "ru"
                                                                 QUERY PLAN

--------------------------------------------------------------------------------------------------------------------------------------------
 Finalize Aggregate  (cost=127503.21..127503.22 rows=1 width=8) (actual
time=494.276..494.276 rows=1 loops=1)
   ->  Gather  (cost=127502.69..127503.20 rows=5 width=8) (actual
time=494.207..494.271 rows=6 loops=1)
         Workers Planned: 5
         Workers Launched: 5
         ->  Partial Aggregate  (cost=126502.69..126502.70 rows=1 width=8)
(actual time=487.290..487.290 rows=1 loops=6)
               ->  Parallel Seq Scan on test  (cost=0.00..121502.55
rows=2000055 width=8) (actual time=0.019..311.045 rows=1666667 loops=6)
 Planning time: 0.122 ms
 Execution time: 509.248 ms
(8 rows)


On Tue, Oct 4, 2016 at 3:38 AM, Nikolay Samokhvalov <samokhvalov@gmail.com>
wrote:

> I have a Postgres 9.6.0 instance for testing purposes. It turned out that
> in postgresql.conf, this postgres cluster has "default_text_search_config"
> being set to "ru" for some reason.
>
> Suddenly, I encountered with an error relating to full text configuration
> in test database, where I didn't use FTS at all, I just tried to test
> parallel aggregation feature:
>
> test=# create table test as
> test-#     select
> test-#         i as id,
> test-#         now() - random() * '5 years'::interval as ts,
> test-#         (random() * 100000000)::int4 as number,
> test-#         repeat('x', (10 + 40 * random())::int4) as txt
> test-#     from
> test-#         generate_series(1, 10000000) i;
> SELECT 10000000
>
> test=# alter table test add primary key (id);
> ALTER TABLE
>
> test=# explain analyze select min(ts) from test;
>                                                        QUERY PLAN
> ------------------------------------------------------------
> -------------------------------------------------------------
>  Aggregate  (cost=226505.45..226505.46 rows=1 width=8) (actual
> time=1460.227..1460.227 rows=1 loops=1)
>    ->  Seq Scan on test  (cost=0.00..201504.76 rows=10000276 width=8)
> (actual time=0.013..697.676 rows=10000000 loops=1)
>  Planning time: 0.098 ms
>  Execution time: 1460.257 ms
> (4 rows)
>
> test=# set max_parallel_workers_per_gather to 8;
> SET
> test=#  explain analyze select min(ts) from test;
> *ERROR:  invalid value for parameter "default_text_search_config": "ru"*
> *CONTEXT:  parallel worker*
>
> Such an error would be OK if I tried to use some text search function
> (since really no 'ru' configuration is present in 'test' database), but I
> ovbviously didn't.
>
> As you can see, no FTS features are involved in table definition, so it
> seems really strange to see FTS-related error
>
> Bottom line, steps to reproduce:
>  1) put not existing text search default configurtion
>  2) set max_parallel_workers_per_gather to 8;
>  3) do some aggregation query that would involved parallel workers and
> doesn't use FTS features.
>
> What I expected: no FTS-related error, because I FTS was not used by the
> query.
>
>