Josh Berkus wrote:
Josh,
Just my own two cents. First I am not knocking the work that has been on
autovacuum. I am sure that it was a leap on its own to get it to work.
However I will say that I just don't see the reason for it.
I've personally seen at least a dozen user requests for
Matthew T. O'Connor wrote:
Joshua D. Drake wrote:
Just my own two cents. First I am not knocking the work that has been
on autovacuum. I am sure that it was a leap on its own to get it to
work. However I will say that I just don't see the reason for it.
The major reasons for autovacuum as
* Reduces the total amount of time the system spends vacuuming since
it only vacuums when needed.
Can be easily done with cron.
* Keeps stats up-to-date automatically
Which can be done with cron
* Eliminates newbie confusion
RTFM
* Eliminates one of the criticisms that the public
Note that issuing warnings due to normal DML SQL queries is much more severe
than the typical DDL warnings. Many people have queries strewn throughout the
application so updating them may be a *lot* of work. And for an app issuing
hundreds or thousands of queries per minute (or even second) a
On Jun 17, 2005, at 4:34 PM, Greg Stark wrote:
And for an app issuing
hundreds or thousands of queries per minute (or even second) a
warning could
effectively be a showstopper. It could require disabling all
warnings in their
config to avoid filling their disk with Postgres logs in
One related idea that I have been meaning to moot for a while
now though, is that of a 'utility' database. One of the
problems we've always had in pgAdmin (and presumably
phpPgAdmin as well), is that the only database we know exists
with any reasonable surety is template1, and
I also think it is useful and make things easier.
A connection on template1 also prevent others to create new databases.
connection1:
template1#=
connection2:
foo=# create database bar;
ERROR: source database template1 is being accessed by other users
---(end of
Qingqing Zhou wrote:
Tom Lane [EMAIL PROTECTED] writes:
Yeah --- a libpq-based solution is not what I think of as integrated at
all, because it cannot do anything that couldn't be done by the existing
external autovacuum process. About all you can buy there is having the
postmaster spawn the
Magnus Hagander wrote:
fer enhanced functionality in the client.
To overcome this, a alternative database created by initdb
would be very useful. This would be roughly the equivalent of
SQL Server's 'msdb'
database and would allow:
- A default non-template database for apps to connect to
I think this is a very good idea. I've come up against this need once or
twice before.. And the fact that stuff in template1 gets propagated out
to all newly created databases can be a major pain when this happens.
A shared database for this stuff would be great - then each tool could
just
Christopher Kings-Lynne wrote:
In phpPgAdmin the default db to connect to can be specified per-server
in the config file. It defaults to template1. It actually is not
relevant at all which db it is, so long as they can connect to it.
I wonder how many users actually change that value for
The major reasons for autovacuum as I see it are as follows:
* Reduces administrative overhead having to keep track of what tables
need to be vacuumed how often.
Creates more overhead and thus reduces performance.
Or reduces vacuum overhead because the vacuum strategy is much better
Added to TODO:
* Create a bitmap of pages that need vacuuming
Instead of sequentially scanning the entire table, have the background
writer or some other process record pages that have expired rows, then
VACUUM can look at just those pages rather than the entire table. In
In phpPgAdmin the default db to connect to can be specified per-server
in the config file. It defaults to template1. It actually is not
relevant at all which db it is, so long as they can connect to it.
I wonder how many users actually change that value for php/pgadmin or
simply leave it
4) Related to this, I guess, is that a user's FSM settings might be
completely inappropriate. The 'Just read the manual' or 'Just read the
logs' argument doesn't cut it, because the main argument for autovacuum in
the backend is that people do not and will not.
Agreed, it doesn't
On Fri, 17 Jun 2005 06:26 pm, Andreas Pflug wrote:
Qingqing Zhou wrote:
Tom Lane [EMAIL PROTECTED] writes:
Yeah --- a libpq-based solution is not what I think of as integrated at
all, because it cannot do anything that couldn't be done by the existing
external autovacuum process. About
On Fri, 17 Jun 2005, Russell Smith wrote:
4) Related to this, I guess, is that a user's FSM settings might be
completely inappropriate. The 'Just read the manual' or 'Just read the
logs' argument doesn't cut it, because the main argument for autovacuum in
the backend is that people do not
On Fri, 17 Jun 2005, Russell Smith wrote:
Added to TODO:
* Create a bitmap of pages that need vacuuming
Instead of sequentially scanning the entire table, have the background
writer or some other process record pages that have expired rows, then
VACUUM can look at just
Christopher Kings-Lynne wrote:
In phpPgAdmin the default db to connect to can be specified
per-server in the config file. It defaults to template1. It
actually is not relevant at all which db it is, so long as they can
connect to it.
I wonder how many users actually change that value for
-BEGIN PGP SIGNED MESSAGE-
Hash: SHA1
Hi,
RPM building for Fedora Core 4 and PostgreSQL 7.4.8 failed with the error
below:
==
make[3]: Leaving directory
`/usr/src/redhat/BUILD/postgresql-7.4.8/src/pl/tcl'
make[3]: Entering
Joshua D. Drake wrote:
Josh Berkus wrote:
I've personally seen at least a dozen user requests for autovacuum
in the backend, and had this conversation about 1,100 times:
NB: After a week, my database got really slow.
Me: How often are you running VACUUM ANALYZE?
NB: Running what?
Can't
Joshua D. Drake wrote:
Matthew T. O'Connor wrote:
The major reasons for autovacuum as I see it are as follows:
* Reduces administrative overhead having to keep track of what tables
need to be vacuumed how often.
Creates more overhead and thus reduces performance.
In the general case,
Russell Smith wrote:
* Reduces the total amount of time the system spends vacuuming since it
only vacuums when needed.
Can be easily done with cron.
Can you do partial table vacuums with CRON?
You can work out the smartest time to vacuum with cron? I thought it just
scheduled
Russell Smith wrote:
On Fri, 17 Jun 2005 06:26 pm, Andreas Pflug wrote:
Qingqing Zhou wrote:
One reason of not using lib-pq is that this one has to wait for the
completion of each vacuum (we don't has async execution in libpq right?),
There *is* async execution in libpq, and
Probably, though the create db issue is a good reason not to use template1.
Create db issue?
So may I propose to have a pg_system database created by initdb, as a
copy from template1 in 8.1?
But then dbas will block off access to that db, or drop it and we're
back to square one...
Chris
Michael Glaesemann wrote:
On Jun 17, 2005, at 4:34 PM, Greg Stark wrote:
And for an app issuing
hundreds or thousands of queries per minute (or even second) a
warning could
effectively be a showstopper. It could require disabling all
warnings in their
config to avoid filling
-Original Message-
From: Christopher Kings-Lynne [mailto:[EMAIL PROTECTED]
Sent: Fri 6/17/2005 9:47 AM
To: Magnus Hagander
Cc: Dave Page; Josh Berkus; pgsql-hackers@postgresql.org
Subject: Re: [HACKERS] Utility database (Was: RE: Autovacuum in the backend)
In phpPgAdmin the default
-Original Message-
From: Christopher Kings-Lynne [mailto:[EMAIL PROTECTED]
Sent: Fri 6/17/2005 11:00 AM
To: Andreas Pflug
Cc: Magnus Hagander; Dave Page; Josh Berkus; pgsql-hackers@postgresql.org; Tom
Lane
Subject: Re: [HACKERS] Utility database (Was: RE: Autovacuum in the backend)
Christopher Kings-Lynne wrote:
Probably, though the create db issue is a good reason not to use
template1.
Create db issue?
CREATE TABLE (implicitely using TEMPLATE template1) often fails because
template1 has connections exceeding the current one.
So may I propose to have a pg_system
On 6/17/05, Josh Berkus josh@agliodbs.com wrote:
Hey, Folks,
I need to find someone who's really interesed in working with DTrace. Sun
has offered to help put DTrace probes into PostgreSQL for advanced
profiling, but need to know where to probe. Anyone?
I'm afraid that I won't get
[EMAIL PROTECTED] (Gavin Sherry) wrote:
I guess the main point is, if something major like this ships in the
backend it says to users that the problem has gone away. pg_autovacuum is
a good contrib style solution: it addresses a problem users have and
attempts to solve it the way other users
Matthew T. O'Connor matthew@zeut.net writes:
... People just didn't like including libpq
into the backend for reasons I don't remember.
One reason I can think of is that there would be global-symbol conflicts
--- libpq has copies of some backend routines, but they are not
identical.
In any
But then dbas will block off access to that db, or drop it
and we're
back to square one...
Don't see why they would. Let's review what we have here:
Database Function(s)
template0 guaranteed-virgin template for CREATE DATABASE
template1
Christopher Kings-Lynne [EMAIL PROTECTED] writes:
So may I propose to have a pg_system database created by initdb, as a
copy from template1 in 8.1?
Seems like a bizarre choice of name. Why not default?
But then dbas will block off access to that db, or drop it and we're
back to square
Magnus Hagander [EMAIL PROTECTED] writes:
It wouldn't just be default to connect to, it would also be location
for tools to store cluster-wide information. Which makes pg_system a
slightly more reasonable name in that context, but i certainly have no
problem with default as a name.
Well,
It wouldn't just be default to connect to, it would also be
location for tools to store cluster-wide information. Which makes
pg_system a slightly more reasonable name in that context, but i
certainly have no problem with default as a name.
Well, where a tool chooses to install stuff
On Fri, 2005-06-17 at 00:03 -0700, Joshua D. Drake wrote:
Matthew T. O'Connor wrote:
Joshua D. Drake wrote:
Just my own two cents. First I am not knocking the work that has been
on autovacuum. I am sure that it was a leap on its own to get it to
work. However I will say that I just
Magnus Hagander wrote:
I dislike the name pg_system because it implies that that DB
is somehow special from the point of view of the system ...
which is exactly what it would *not* be.
That I can certainly agree with.
I suggested the name to indicate that it's a db used by system
Tom Lane wrote:
One argument against this is that it'd mean another copy of the system
catalogs in a standard installation. That's been running three to five
megabytes over the last few releases. Disk space is pretty cheap these
days, but we do get occasional complaints from people who wish
Josh,
Just so everyone knows from the get go here. I am purposely playing a
little devils advocate.
Well, please stop it. We discussed AV over a year ago when we ran out of time
to integrate it with 8.0. This disucussion now is hindering any discussion
of what needs to be *done* to
Christopher Browne wrote:
[EMAIL PROTECTED] (Gavin Sherry) wrote:
I guess the main point is, if something major like this ships in the
backend it says to users that the problem has gone away. pg_autovacuum is
a good contrib style solution: it addresses a problem users have and
attempts to
Andreas Pflug [EMAIL PROTECTED] writes:
I particularly dislike the name default for that database, because
we'd have to expect users to place their user data there regularly (as
in the public schema), which is just what should *not* happen.
Why not?
Any tools using this database for their
Tom Lane wrote:
Andreas Pflug [EMAIL PROTECTED] writes:
I particularly dislike the name default for that database, because
we'd have to expect users to place their user data there regularly (as
in the public schema), which is just what should *not* happen.
Why not?
Any tools using this
With libreadline, we are not taking their code or distributing it, but
merely linking to it if it exists.
But we are also requiring it. The rpms won't install unless readline is
available.
Now, some say that is enough to make
us GPL, but many don't agree with that interpretation.
On Fri, 17 Jun 2005, Joshua D. Drake wrote:
With libreadline, we are not taking their code or distributing it, but
merely linking to it if it exists.
But we are also requiring it. The rpms won't install unless readline is
available.
that isn't a PostgreSQL requirement though, that is a
Am Dienstag, den 14.06.2005, 22:59 -0300 schrieb Marc G. Fournier:
We already do ... libreadline ...
Hm. I remember in my source builds I used libedit
which is the BSD replacement IIRC?
---(end of broadcast)---
TIP 1: subscribe and unsubscribe
Joshua D. Drake wrote:
With libreadline, we are not taking their code or distributing it, but
merely linking to it if it exists.
But we are also requiring it. The rpms won't install unless readline
is available.
Now, some say that is enough to make
us GPL, but many don't agree
Marc G. Fournier wrote:
On Fri, 17 Jun 2005, Joshua D. Drake wrote:
With libreadline, we are not taking their code or distributing it, but
merely linking to it if it exists.
But we are also requiring it. The rpms won't install unless readline
is available.
that isn't a PostgreSQL
Huh ?
./configure --without-readliine
works just fine, there is no requirement.
Dave
On 17-Jun-05, at 3:04 PM, Joshua D. Drake wrote:
Marc G. Fournier wrote:
On Fri, 17 Jun 2005, Joshua D. Drake wrote:
With libreadline, we are not taking their code or distributing
it, but
merely
Joshua D. Drake wrote:
Dave Cramer wrote:
Huh ?
./configure --without-readliine
works just fine, there is no requirement.
Again:
If we **link** to readline, postgresql won't start without it.
That is a postgresql requirement. Yes we can compile without
it. That isn't what I
Joshua D. Drake wrote:
If we link to readline, postgresql won't start without it. Regardless
of the package. That seems pretty much a postgresql requirement ;)
If you think you're in danger don't link to it. You don't have to at
all. You can build without readline entirely (it's only
Dave Cramer wrote:
Huh ?
./configure --without-readliine
works just fine, there is no requirement.
Again:
If we **link** to readline, postgresql won't start without it.
That is a postgresql requirement. Yes we can compile without
it. That isn't what I was talking about.
But as Andrew
-Original Message-
From: Tom Lane [mailto:[EMAIL PROTECTED]
Sent: 17 June 2005 15:09
To: Christopher Kings-Lynne
Cc: Andreas Pflug; Magnus Hagander; Dave Page; Josh Berkus;
pgsql-hackers@postgresql.org
Subject: Re: [HACKERS] Utility database (Was: RE: Autovacuum
in the
Martha Stewart called it a Good Thing when [EMAIL PROTECTED] (Andreas Pflug)
wrote:
Magnus Hagander wrote:
I dislike the name pg_system because it implies that that DB is
somehow special from the point of view of the system ... which is
exactly what it would *not* be.
That I can certainly
In the last exciting episode, dpage@vale-housing.co.uk (Dave Page) wrote:
But then dbas will block off access to that db, or drop it and
we're back to square one...
That's their choice though, and it would then be up to them to
provide an alternative for their users (there's nothing to stop
-Original Message-
From: Andreas Pflug [mailto:[EMAIL PROTECTED]
Sent: 17 June 2005 18:45
To: Tom Lane
Cc: Christopher Kings-Lynne; Magnus Hagander; Dave Page; Josh
Berkus; pgsql-hackers@postgresql.org
Subject: Re: [HACKERS] Utility database (Was: RE: Autovacuum
in the
Title: MemoryContextAlloc: invalid request size
I got the copy of the database and ran it
with truss.
From the trace log it looked like ${PGDATA/}base/last-number/pg_internal.init
was corrupted
I replaced this file with ${PGDATA/}base/prev-number/pg_internal.init
After that I was
-Original Message-
From: [EMAIL PROTECTED]
[mailto:[EMAIL PROTECTED] On Behalf Of
Christopher Browne
Sent: 17 June 2005 19:59
To: pgsql-hackers@postgresql.org
Subject: Re: [HACKERS] Utility database
Thus, sys_shared, def_share, user_commons are all sorts of names
that
Dave Page wrote:
Thus, sys_shared, def_share, user_commons are all sorts of names
that suggest that this is some sort of default/shared area.
I like the first. The second and third seem less obvious to me.
'default_shared' should definitely get the point across, though it's a
little
Andrew Dunstan wrote:
It strikes me that these names just might have some significance to
developers but will have none at all for users. I don't heve a better
alternative ... maybe because the purpose has been expressed somewhat
fuzzily.
I'd define the purpose like this:
- being a db
On Fri, Jun 17, 2005 at 12:21:44PM -0400, Matthew T. O'Connor wrote:
snip more stuff about how we need to track pages with dead tuples
This will make VACUUM less painful, but it doesn't eliminate the need /
desire for autovacuum. I agree this would be good, but I see it as a
separate
Joshua D. Drake [EMAIL PROTECTED] writes:
With libreadline, we are not taking their code or distributing it, but
merely linking to it if it exists.
But we are also requiring it. The rpms won't install unless readline is
available.
The RPMs require it --- not our source code. Since the
Dave Page dpage@vale-housing.co.uk writes:
Personally I prefer the first or last, as default implies to me that
it's a kindof general use database - which, as Tom points out it could
be, however I think it's better to encourage users to only use it as
directed by tool providers, and not for
Dave Page dpage@vale-housing.co.uk writes:
I like the first. The second and third seem less obvious to me.
'default_shared' should definitely get the point across, though it's a
little long.
I think shared would give the wrong impression to many people ---
nowadays the connotation of that is
On 6/18/05, Tom Lane [EMAIL PROTECTED] wrote:
What is important is that it is possible, and useful, to build Postgres
in a completely non-GPL environment. If that were not so then I think
we'd have some license issues. But the fact that building PG in a
GPL-ized environment creates a
65 matches
Mail list logo