[postgres-xl-bugs] Strange error

asvany asvany at gmail.com
Fri Sep 29 07:53:30 PDT 2017


Hi,  

It was my fault.

I used different identifiers for the data nodes in the pgxc_ctl.conf and
the data nodes' postgresql.conf.  

By the way, some warning message maybe could help here. :)  

,Asvany


On 08/07/2017 01:10 PM, Tomas Vondra wrote:
> Hi,
>
> On 08/03/2017 05:04 PM, asvany wrote:
>> Hi Devs,
>>
>> My database start behaving in a very strange way.
>>
>> I except this result
>> 1
>> 2
>> 3
>> 4
>>
>> from this query:
>>
>>    CREATE SCHEMA IF NOT EXISTS temp;
>>      DROP TABLE IF EXISTS temp.asd;
>>
>>      CREATE TABLE IF NOT EXISTS temp.asd AS
>>                           SELECT 1 as c;
>>
>>      INSERT INTO temp.asd SELECT 2 as c;
>>
>>      INSERT INTO temp.asd SELECT 3 as c LIMIT 10;
>>
>>      INSERT INTO temp.asd SELECT 4 as c GROUP BY c;
>>
>>      SELECT * FROM temp.asd ORDER BY c asc;
>>
>> but produces this result :
>>
>> 1
>> 2
>>
>>
>>
>> because this lines INSERT INTO temp.asd SELECT 3 as c LIMIT 10; notices:
>> /Query returned successfully: 0 rows affected, 121 msec execution time.
>>
>> /It don't insert anything.
>>
>
> Where does this notice come from? A datanode or a coordinator?
>
>> query plan:
>> "Remote Subquery Scan on all
>> (data_pg00,data_pg01,data_pg02,data_pg03,data_pg04,data_pg05,data_pg06,data_pg07)
>> (cost=0.00..0.02 rows=1 width=4)"
>> "  ->  Insert on asd  (cost=0.00..0.02 rows=1 width=4)"
>> "        ->  Remote Subquery Scan on all (data_pg00) (cost=0.00..0.02
>> rows=1 width=4)"
>> "              Distribute results by H: c"
>> "              ->  Limit  (cost=0.00..0.01 rows=1 width=0)"
>> "                    ->  Result  (cost=0.00..0.01 rows=1 width=0)"
>>
>>
>> No trace of any error in the log files.
>>
>> (I tried the latest source , restart , reboot nodes )
>>
>> I create a new cluster instance with the same nodes with the same
>> binaries and configuration files and the query works well, produces
>> the expected result !
>> It produces the same query plan.
>>
>> I made a new database in the original server and as I see the error
>> is not database dependent , so it affects the whole server.
>>
>>
>
> This seems like some sort of cluster consistency issue. Can you check
> that pgxc_node contents is the same on all nodes?
>
> What version of Postgres-XL is this, actually?
>
>
> regards
>



More information about the postgres-xl-bugs mailing list