Re: [PERFORM] Strange behavior: pgbench and new Linux kernels

2009-03-31 Thread Greg Smith
On Tue, 31 Mar 2009, Kevin Grittner wrote: On Thu, Apr 17, 2008 at 7:26 PM, Greg Smith wrote: On this benchmark 2.6.25 is the worst kernel yet: It seems I have a lot of work ahead of me here to nail down and report what's going on here. I don't remember seeing a follow-up on this issue

Re: [PERFORM] How to get parallel restore in PG 8.4 to work?

2009-03-31 Thread Tom Lane
henk de wit writes: > For performance reasons (obviously ;)) I'm experimenting with parallel > restore in PG 8.4. I grabbed the latest source snapshot (of today, March 30) > and compiled this with zlib support. I dumped a DB from PG 8.3.5 (using > maximum compression). I got this message howeve

[PERFORM] How to get parallel restore in PG 8.4 to work?

2009-03-31 Thread henk de wit
Hi, For performance reasons (obviously ;)) I'm experimenting with parallel restore in PG 8.4. I grabbed the latest source snapshot (of today, March 30) and compiled this with zlib support. I dumped a DB from PG 8.3.5 (using maximum compression). I got this message however: postg...@mymachine:/h

Re: [PERFORM] Strange behavior: pgbench and new Linux kernels

2009-03-31 Thread Kevin Grittner
>>> On Thu, Apr 17, 2008 at 7:26 PM, Greg Smith wrote: > On this benchmark 2.6.25 is the worst kernel yet: > It seems I have a lot of work ahead of me here > to nail down and report what's going on here. I don't remember seeing a follow-up on this issue from last year. Are there still any p

Re: [PERFORM] Very specialised query

2009-03-31 Thread Віталій Тимчишин
> > > The outer nested join has the VALUES as the main loop, and the complicated > join as the leaf. So, the complicated overlap-finding join gets run twice. That's weird. What do you have as statistics target? Planner is incorrect few orders of magnitude, so increasing it may help. BTW: One of c

Re: [PERFORM] Very specialised query

2009-03-31 Thread Matthew Wakeling
On Mon, 30 Mar 2009, Віталій Тимчишин wrote: select case when n == 1 then id1 else id2 end, case when n == 2 then id1 else id2 end from ( SELECT    l1.id AS id1,    l2.id AS id2 FROM    location l1,    location l2 WHERE        l1.objectid = 22893    AND l2.objectid = 22893    AND l1.id <

Re: [PERFORM] I have a fusion IO drive available for testing

2009-03-31 Thread Chris Browne
craig_ja...@emolecules.com (Craig James) writes: > Dave Cramer wrote: >> So I tried writing directly to the device, gets around 250MB/s, >> reads at around 500MB/s >> >> The client is using redhat so xfs is not an option. > > I'm using Red Hat and XFS, and have been for years. Why is XFS not an opt

Re: [PERFORM] I have a fusion IO drive available for testing

2009-03-31 Thread Craig James
Dave Cramer wrote: So I tried writing directly to the device, gets around 250MB/s, reads at around 500MB/s The client is using redhat so xfs is not an option. I'm using Red Hat and XFS, and have been for years. Why is XFS not an option with Red Hat? Craig -- Sent via pgsql-performance mail

Re: [PERFORM] I have a fusion IO drive available for testing

2009-03-31 Thread Dave Cramer
On Fri, Mar 27, 2009 at 4:33 PM, David Rees wrote: > On Fri, Mar 27, 2009 at 10:30 AM, wrote: > > On Thu, 26 Mar 2009, Dave Cramer wrote: > >> So far using dd I am seeing around 264MB/s on ext3, 335MB/s on ext2 > write > >> speed. So the question becomes what is the best filesystem for this > d

Re: [PERFORM] Very specialised query

2009-03-31 Thread Matthew Wakeling
On Mon, 30 Mar 2009, Marc Mamin wrote: Are your objects limited to some smaller ranges of your whole interval ? If yes you may possibly reduce the ranges to search for while using an additional table with the min(start) max(end) of each object... No, they aren't. However, even if they were, th