Dear all,
I'm running a query from Java on a postgres database:
Java version: 17
JDBC version: 42.4.2
Postgres version: 13.1
In parallel I'm testing the same queries from pgAdmin 4 version 6.13
The tables I'm using contains more than 10million rows each and I have two
questions here:
1. I nee
You are right but the correct query is the one in the attached file. What
we want to do here is to force psql to send the "wrong data types" to
postgres and, as a result of this, get a bad plan.
Cheers,
Jose Osinde
On Thu, Aug 25, 2022 at 3:48 PM Ranier Vilela wrote:
> Em qui.,
d of
select the right one and uses sec scans for all the cases.
Attached the result.
Again, many thanks for your help,
Jose Osinde
explain analyze select logical_identifier, version_id, lastproduct
FROM test_product_ui_partition.product_ui pui
WHERE
pui.mission_id='urn:esa:psa:cont
-20220729t235959'::text))
Rows Removed by Filter: 33
Planning Time: 16.874 ms
Execution Time: .496 ms
Elapsed: 10122 ms
Source code of the test is attached in the mail. It provides some other
test cases as:
- Query other non partitioned table, with correct performance
- Query the correct partitioned table according to the partition key
column (mission_id). In this case only that table is scanned, but again,
using PARALLEL SEQ SCAN and not the INDEX SCAN
Could you please provide any hint on the possible reasons of this behavior
and the performance degradation that is affecting only the JAVA client.
Best regards,
Jose Osinde
JdbcTest.java
Description: Binary data
