Re: [GENERAL] Problem with update on partitioned table

2008-03-24 Thread Martin Gainty
= on HTH Martin - Original Message - From: "Alex Solovey" <[EMAIL PROTECTED]> To: Sent: Monday, March 24, 2008 2:15 PM Subject: [GENERAL] Problem with update on partitioned table > Hello, > > We have pretty big production database (running PostgreSQL 8.3.1) with

Re: [GENERAL] Problem with update on partitioned table

2008-03-24 Thread Tom Lane
Alex Solovey <[EMAIL PROTECTED]> writes: > Is this patch going to be included in 8.3 only, or in 8.2 as well? I thought I was already taking a chance by putting it in REL8_3_STABLE. Since 8.2 doesn't have a regression compared to the previous release, and is also very much more different from HEAD

Re: [GENERAL] Problem with update on partitioned table

2008-03-24 Thread Alex Solovey
Tom, > If you're feeling brave, try the patch I just committed to CVS. I just did it. It works! According to the query plan, only one partition is being examined now. Is this patch going to be included in 8.3 only, or in 8.2 as well? Thanks! - Sent via pgsql-general mailing list (pgsql-gene

Re: [GENERAL] Problem with update on partitioned table

2008-03-24 Thread Tom Lane
Alex Solovey <[EMAIL PROTECTED]> writes: > Thanks for the patch. We've tried it here, and it improved query plan > slightly (indeed, it looks exactly like the plan from 8.2.6 now). > But, as you've said, empty sub-joins are still not discarded, so query > execution time did not improve. And this

Re: [GENERAL] Problem with update on partitioned table

2008-03-24 Thread Alex Solovey
Tom, Thanks for the patch. We've tried it here, and it improved query plan slightly (indeed, it looks exactly like the plan from 8.2.6 now). But, as you've said, empty sub-joins are still not discarded, so query execution time did not improve. And this is the same in both 8.2 and 8.3. Note th

Re: [GENERAL] Problem with update on partitioned table

2008-03-24 Thread Alex Solovey
Martin, > which if you dont want to scan ALL partitions must be set to 'on' > constraint_exclusion = on It is 'ON'. The problem is that it does not work well for 'UPDATE foo ... FROM bar' queries, when partitioned table 'foo' is joined with another table. Martin Gainty wrote: Alex- http://

Re: [GENERAL] Problem with update on partitioned table

2008-03-24 Thread Tom Lane
Alex Solovey <[EMAIL PROTECTED]> writes: > We have pretty big production database (running PostgreSQL 8.3.1) with > many partitioned tables. In most cases, they work well (since 8.2.1 at > least) -- constraint exclusion is able to select correct partitions. > However, there is an exception: quer

[GENERAL] Problem with update on partitioned table

2008-03-24 Thread Alex Solovey
Hello, We have pretty big production database (running PostgreSQL 8.3.1) with many partitioned tables. In most cases, they work well (since 8.2.1 at least) -- constraint exclusion is able to select correct partitions. However, there is an exception: queries on partitioned tables using Postgre