It's not an easy decision. Linux isn't wrong. Solaris isn't wrong.
Most people never hit these problems, and the people that do, are
just as likely to hit one problem, or the other. The grass is always
greener on the side of the fence that isn't hurting me right now,
and all that.
Cheers,
On Mon, 3 Oct 2005, Jeff Davis wrote:
involved, but I could be wrong. Is it possible to be hit by the OOM
killer if no applications use fork()?
Sure, whenever the system is out of mem and the os can't find a free page
then it kills a process. If you check the kernel log you can see if the
On Tue, 2005-10-04 at 00:21 -0400, Tom Lane wrote:
Simon Riggs [EMAIL PROTECTED] writes:
AFAICS the following opportunities exist, without changing any of the
theoretical algorithms or the flexibility of definable datatypes:
1. tuplesort_heap_siftup and tuplesort_heap_insert make no
On Mon, Oct 03, 2005 at 11:47:57PM -0700, Jeff Davis wrote:
I think that I've run into the OOM killer without a fork() being
involved, but I could be wrong. Is it possible to be hit by the OOM
killer if no applications use fork()?
fork() is the obvious overcomitter. If Netscape wants to spawn
On Tue, Oct 04, 2005 at 12:37:51AM +0100, Simon Riggs wrote:
On Mon, 2005-10-03 at 23:25 +0200, Martijn van Oosterhout wrote:
Please note: if inlineApplySortFunction() is actually inlined (it isn't
by default)
Can you explain your last post some more. Thats not what I get.
The inline
On Mon, Oct 03, 2005 at 10:51:32PM +0100, Simon Riggs wrote:
Basically, I recommend adding -Winline -finline-limit-1500 to the
default build while we discuss other options.
I add -Winline but get no warnings. Why would I use -finline-limit-1500?
I'm interested, but uncertain as to what
On Tue, 2005-10-04 at 00:26 -0400, Rod Taylor wrote:
As I understand it vacuum operates outside of the regular transaction
and if you stop it (SIGTERM, or pg_cancel_backend()) some of the work it
accomplished will be kept when it rolls back.
For large structures with a ton of dead entries
On Tue, Oct 04, 2005 at 11:55:58AM +0200, Martijn van Oosterhout wrote:
It goes in to say that the limit is 1 for gcc 2.95, but if you
examine the manual for gcc 3.3 it has the limit at 600. So it's
entirely possible that at the time the person wrote that code, it *was*
being inlined, but
I have some doubts about the new pg_config behavior of printing not recorded
when an item is not known. I doubt that this will cause automated users of
this program to behave reasonably in error scenarios. I think a more
reasonable behavior would be to use a non-zero exit status and possibly
Hello hackers,
I'm fiddling around with the backend and have created a memory bug. I
guess I'm overriding a palloced chunk, but can't figure out where.
The backend's report is:
WARNING: problem in alloc set MessageContext: req size alloc size for
chunk 0x8383ca8 in block 0x83834a0
On Tue, 2005-10-04 at 12:37 +0200, Martijn van Oosterhout wrote:
Since we're pretty much relying on gcc to inline for performance, I
still think we should add -Winline by default so we can tell when it's
not doing what we want.
Very much agree to this. Can we put that in the build for 8.1
On Tue, 2005-10-04 at 12:04 +0200, Martijn van Oosterhout wrote:
On Mon, Oct 03, 2005 at 10:51:32PM +0100, Simon Riggs wrote:
Basically, I recommend adding -Winline -finline-limit-1500 to the
default build while we discuss other options.
I add -Winline but get no warnings. Why would I
On Tue, Oct 04, 2005 at 01:11:41PM +0200, Markus Schiltknecht wrote:
Hello hackers,
I'm fiddling around with the backend and have created a memory bug. I
guess I'm overriding a palloced chunk, but can't figure out where.
There are some defines (MEMORY_CONTEXT_CHECKING and
On Tue, Oct 04, 2005 at 12:24:54PM +0100, Simon Riggs wrote:
How did you determine the 1500 figure? Can you give some more info to
surround that recommendation to allow everybody to evaluate it?
[EMAIL PROTECTED]:~/dl/cvs/pgsql-local/src/backend/utils/sort$ gcc
-finline-limit-1000 -Winline -O2
-Original Message-
From: [EMAIL PROTECTED]
[mailto:[EMAIL PROTECTED] On Behalf Of Marc
G. Fournier
Sent: 04 October 2005 02:34
To: pgsql-hackers@postgresql.org
Subject: [HACKERS] New Point Releases Available
Just bundled up 7.3.11, 7.4.9 and 8.0.4 ... please look them
Peter Eisentraut [EMAIL PROTECTED] writes:
I have some doubts about the new pg_config behavior of printing not
recorded
when an item is not known. I doubt that this will cause automated users of
this program to behave reasonably in error scenarios. I think a more
reasonable behavior
The constants related to inlining involve pcode, not actual assembly
instructions, and are compiler version dependent as well as subject to change
without notice by the GNU folks...
from:
http://gcc.gnu.org/onlinedocs/gcc-3.3.5/gcc/Optimize-Options.html#Optimize-Options
-finline-limit=n
By
Martijn van Oosterhout kleptog@svana.org writes:
On Tue, Oct 04, 2005 at 01:11:41PM +0200, Markus Schiltknecht wrote:
I'm fiddling around with the backend and have created a memory bug. I
guess I'm overriding a palloced chunk, but can't figure out where.
There are some defines
On Sat, Oct 01, 2005 at 02:00:12PM -0400, Tom Lane wrote:
I wrote:
It's presumably mostly in the pg_table_is_visible() calls.
I did some profiling on a test case with 1 tables, and noticed that
a big part of the problem is that the catalog caches become entirely
useless: almost every
Martijn van Oosterhout kleptog@svana.org writes:
A quick binary search puts the cutoff between 1200 and 1300. Given
version variation I picked a nice round number, 1500.
Ugh, that's for -O2, for -O3 and above it needs to be 4100 to work.
Maybe we should go for 5000 or so.
I'm using: gcc
On Mon, Oct 03, 2005 at 12:11:49AM -0400, Tom Lane wrote:
Bruce Momjian pgman@candle.pha.pa.us writes:
Jim C. Nasby wrote:
Watching the discussion about how to handle nextval() and keep it dump
compatible makes me wonder: would it be useful to encode database or
dump version info into
On Tue, 2005-10-04 at 16:30 +0200, Martijn van Oosterhout wrote:
On Tue, Oct 04, 2005 at 10:06:24AM -0400, Tom Lane wrote:
Martijn van Oosterhout kleptog@svana.org writes:
I'm using: gcc (GCC) 3.3.5 (Debian 1:3.3.5-13)
I don't know what the units of this number are, but it's apparently
Martijn van Oosterhout kleptog@svana.org writes:
1. Add -Winline so we can at least be aware of when it's (not) happening.
Yeah, I agree with that part, just not with adding a fixed -finline-limit
value.
While on the subject of gcc warnings ... if I touch that code, I want to
remove
Is there enough for a TODO here?
On Mon, Oct 03, 2005 at 11:24:30PM -0400, Greg Stark wrote:
Hannu Krosing [EMAIL PROTECTED] writes:
On P, 2005-10-02 at 23:00 -0400, Tom Lane wrote:
Here's another interesting case to think about:
ALTER TABLE ADD foo integer DEFAULT 1
Is it reasonable to cancel and restart the vacuum process periodically
No.
How big is that table, anyway? Are you trying a VACUUM FULL, or plain
vacuum?
It's only about 60GB in size but appears it has been missed for nearly a
month for vacuum and probably has a large percentage dead
On Sun, Oct 02, 2005 at 09:46:07PM -0400, Tom Lane wrote:
Fredrik Olsson [EMAIL PROTECTED] writes:
To allow indexes to be inherited so unique, foreign keys and such works
properly with inheritance has been on the todo for quite some time. I
thought that most probably it is a very non
On Tue, Oct 04, 2005 at 10:06:24AM -0400, Tom Lane wrote:
Martijn van Oosterhout kleptog@svana.org writes:
I'm using: gcc (GCC) 3.3.5 (Debian 1:3.3.5-13)
I don't know what the units of this number are, but it's apparently far
too gcc-version-dependent to consider putting into our build
Rod Taylor [EMAIL PROTECTED] writes:
As I understand it vacuum operates outside of the regular transaction
It's a perfectly normal transaction.
Is it reasonable to cancel and restart the vacuum process periodically
No.
How big is that table, anyway? Are you trying a VACUUM FULL, or plain
On Sat, Oct 01, 2005 at 07:59:11PM +0100, Simon Riggs wrote:
On Sat, 2005-10-01 at 10:57 -0500, Jim C. Nasby wrote:
To clarify, this is a hard-coded implementation of what I'm asking for:
http://cvs.distributed.net/viewcvs.cgi/stats-sql/logdb/ in a nutshell:
CREATE TABLE log_other (
On Tue, Oct 04, 2005 at 03:56:53PM +0100, Simon Riggs wrote:
I've been using gcc 3.4 and saw no warning when using either -Winline
or -O3 -Winline.
Ok, I've just installed 3.4 and verified that. I examined the asm code
and gcc is inlining it. I concede, at this point just throw in -Winline
and
Rod Taylor [EMAIL PROTECTED] writes:
The catch is that there are some other very active structures (like
pg_listener for Slony) which after a couple of hours without vacuuming
will quickly have the DB at an unreasonably high load (low tens) which
seems to all but halt the vacuum on the large
Jim C. Nasby [EMAIL PROTECTED] writes:
On Sun, Oct 02, 2005 at 09:46:07PM -0400, Tom Lane wrote:
1. A cross-table index would need to store a table OID as well as the
existing block/offset information in order to tell you what an entry is
pointing at.
Wouldn't it make more sense to use a
Another possibility is optimizing for the special case of
indexing on a partitioning key. In this case, index values
would be very localized to one table, so just storing the
table info on each index page (or something similar) would work well.
If you have the partitioning key in the
On Tue, Oct 04, 2005 at 11:05:49AM -0400, Tom Lane wrote:
Jim C. Nasby [EMAIL PROTECTED] writes:
On Sun, Oct 02, 2005 at 09:46:07PM -0400, Tom Lane wrote:
1. A cross-table index would need to store a table OID as well as the
existing block/offset information in order to tell you what an
Well, I never said unique, but you're correct, it's pretty undesirable
to put a global index on your partitioning key.
On Tue, Oct 04, 2005 at 06:16:21PM +0200, Zeugswetter Andreas DAZ SD wrote:
Another possibility is optimizing for the special case of
indexing on a partitioning key. In
On Tue, Oct 04, 2005 at 05:23:41PM +0200, Martijn van Oosterhout wrote:
On Tue, Oct 04, 2005 at 03:56:53PM +0100, Simon Riggs wrote:
I've been using gcc 3.4 and saw no warning when using either -Winline
or -O3 -Winline.
Ok, I've just installed 3.4 and verified that. I examined the asm code
On Sat, Oct 01, 2005 at 12:51:08PM -0700, Roger Hand wrote:
-Original Message-
From: [EMAIL PROTECTED]
[mailto:[EMAIL PROTECTED] Behalf Of Jim C. Nasby
Sent: Friday, September 30, 2005 4:49 PM
Subject: Re: [PERFORM] [HACKERS] Query in SQL statement
I suggest ditching the
On Tue, 2005-10-04 at 18:16 +0200, Zeugswetter Andreas DAZ SD wrote:
Another possibility is optimizing for the special case of
indexing on a partitioning key. In this case, index values
would be very localized to one table, so just storing the
table info on each index page (or something
This is to announce the first, Alpha, release of Veil, a database
security add-on for Postgres. It allows databases to be created with
efficient row-level access controls; different users have access to
different subsets of data.
Veil is hosted on pgfoundry at http://veil.projects.postgresql.org/
Marc Munro [EMAIL PROTECTED] writes:
Since I was unable to dynamically assign a LWLock using
LWLockAssign (none available), I have fairly arbitrarily overloaded the
use of existing LWLocks. When the flames die down perhaps we can
discuss making a small number (one would be enough for me) of
I am not sure if this idea was mentioned before.
The basic prefix btree idea is quite straightforward, i.e., try to
compress the key items within a data page by sharing the common prefix.
Thus the fanout of the page is increased and the benefits is obvious
theorectically.
Consider the following
41 matches
Mail list logo