Thread: BUG #6482: Service terminates executing "With ... as ... UPDATE" query
BUG #6482: Service terminates executing "With ... as ... UPDATE" query
From
dzolotarevskiy@gmail.com
Date:
The following bug has been logged on the website: Bug reference: 6482 Logged by: Dmitriy Email address: dzolotarevskiy@gmail.com PostgreSQL version: 9.1.2 Operating system: Windows Server 2003 r2 32bit Description:=20=20=20=20=20=20=20=20 All Postgres processes terinates. In log you can find only=20 LOG: server process (PID ...) was terminated by exception OxC0000005. All queries works fine when there is only one session. When Sessions more then 20 it starts terminating with some probability.=20 On my production server it stops about 50 times per day. On my dev pc, with testing utility that uses 50 sessions at one time, it falls down in about 1 minute. Investigating problem i have found the "fatal" query and excluding part by part found out that new syntax produce that exception. To reproduce create 3 tables and execute my query in at least 20 simultaneous threads, each one should use separate session. Tables: user_task (pk id, fk user_proxy_id) user_proxy (pk id, fk proxy_last_id, date) proxy_last (pk id, last_used) Query: with mup as (=20 UPDATE user_proxy=20 SET date =3D now()=20 WHERE id=3D (select user_proxy_id from user_task=20 where id=3D420 )=20 returning proxy_last_id ) UPDATE proxy_last=20 SET last_used =3D now()=20 WHERE id=3D(select proxy_last_id=20 from mup); I have solved the problem rewriting query to old syntax.
dzolotarevskiy@gmail.com writes: > All queries works fine when there is only one session. When Sessions more > then 20 it starts terminating with some probability. > On my production server it stops about 50 times per day. > On my dev pc, with testing utility that uses 50 sessions at one time, it > falls down in about 1 minute. > Investigating problem i have found the "fatal" query and excluding part by > part found out that new syntax produce that exception. I believe this is an already-known issue: http://archives.postgresql.org/pgsql-bugs/2012-01/msg00157.php It will be fixed in next week's update releases. regards, tom lane