Postgres Pro
Downloads
Home   >   mailing lists

Thread: Upgrade to 8.3.0?

Upgrade to 8.3.0?

From
"Shane Wright"
Date:

Hi,

 

I’m considering an upgrade to 8.3.0 for some internal databases – would normally wait until at least .1 release of anything for safety but there’s a lot of nice sounding stuff here!

 

After trawling the bug logs and doing some testing on some servers here all looks ok except for this bug around VACUUM FULL and HOT tables:

http://archives.postgresql.org/pgsql-bugs/2008-02/msg00068.php

 

So my questions are thus...

 

-          Is the HOT/VACUUMFULL bug above dangerous or is its biggest side-effect an error until a non-full vacuum is run?

 

-          Has anyone spotted anyone else that would discourage an upgrade at this stage?

 

If it helps, our use of SQL is fairly basic – nothing fancy, and nothing Postgres-specific.

 

Any help appreciated,

 

Thanks

 

Shane

 

Re: Upgrade to 8.3.0?

From
Tom Lane
Date:
"Shane Wright" <shane.wright@edigitalresearch.com> writes:
> -          Is the HOT/VACUUMFULL bug above dangerous or is its biggest
> side-effect an error until a non-full vacuum is run?

As long as you don't build with --enable-cassert, that bug isn't real
nasty.  Especially if you don't do VACUUM FULL routinely ;-)

            regards, tom lane

Re: Upgrade to 8.3.0?

From
"Shane Wright"
Date:
Tom
 
Many thanks :)
 
I've tried to find this out for myself but have failed :(
 
I'm assuming that the default RPMs for RHEL 4 (on ftp.postgresql.org) are not built with --enable-cassert - and that I can veryify this by using the same reproduction case you demonstrated in the bug history?
 
(I should just get the memory allocation error on attempting to vacuum full - instead of an assertion failing and bringing down the whole PostgreSQL server (panic?))
 
Please could you confirm my assertion (sic) is correct?
 
(I get this behaviour on my test server and all appears to be ok)
 
Kind regards
 
Shane
 

From: Tom Lane [mailto:tgl@sss.pgh.pa.us]
Sent: Fri 29/02/2008 19:46
To: Shane Wright
Cc: pgsql-general@postgresql.org
Subject: Re: [GENERAL] Upgrade to 8.3.0?

Shane Wright writes:
> -          Is the HOT/VACUUMFULL bug above dangerous or is its biggest
> side-effect an error until a non-full vacuum is run?

As long as you don't build with --enable-cassert, that bug isn't real
nasty.  Especially if you don't do VACUUM FULL routinely ;-)

                        regards, tom lane

Re: Upgrade to 8.3.0?

From
Tom Lane
Date:
"Shane Wright" <shane.wright@edigitalresearch.com> writes:
> I'm assuming that the default RPMs for RHEL 4 (on ftp.postgresql.org) are not built with --enable-cassert - and that
Ican veryify this by using the same reproduction case you demonstrated in the bug history? 

I believe they aren't, but you could check by seeing what pg_config
reports as the configure options; or even more directly by seeing if
it will let you "SET debug_assertions TO on".

            regards, tom lane

Re: Upgrade to 8.3.0?

From
"Shane Wright"
Date:
 
hmm - I don't suppose the RHEL builds already include your patch against this issue?  If so I guess it would be useful to know :)
 
I've run through the steps you outlined to reproduce the fault on a new box (after having reproduced it successfully on test servers), and it does not show the problem:
 
 
[root@XXXXXXX data]# !psql
psql test -U XXXXX
Welcome to psql 8.3.0, the PostgreSQL interactive terminal.
Type:  \copyright for distribution terms
       \h for help with SQL commands
       \? for help with psql commands
       \g or terminate with semicolon to execute query
       \q to quit
test=# create table foo as select null::int as x from generate_series(1,2000) x;
SELECT
test=# delete from foo;
DELETE 2000
test=# select count(*) from foo;
 count
-------
     0
(1 row)
test=# vacuum full foo;
VACUUM
test=#
test=# set debug_assertions to on;
ERROR:  assertion checking is not supported by this build
test=#
 
The RPMs installed from are these:
 
postgresql-8.3.0-1PGDG.rhel4
postgresql-libs-8.3.0-1PGDG.rhel4
postgresql-server-8.3.0-1PGDG.rhel4
Thanks
 
Shane

From: Tom Lane [mailto:tgl@sss.pgh.pa.us]
Sent: Fri 29/02/2008 22:55
To: Shane Wright
Cc: pgsql-general@postgresql.org
Subject: Re: [GENERAL] Upgrade to 8.3.0?

"Shane Wright" <shane.wright@edigitalresearch.com> writes:
> I'm assuming that the default RPMs for RHEL 4 (on ftp.postgresql.org) are not built with --enable-cassert - and that I can veryify this by using the same reproduction case you demonstrated in the bug history?

I believe they aren't, but you could check by seeing what pg_config
reports as the configure options; or even more directly by seeing if
it will let you "SET debug_assertions TO on".

                        regards, tom lane

Re: Upgrade to 8.3.0?

From
"Joshua D. Drake"
Date:
On Sat, 1 Mar 2008 03:27:50 -0000
"Shane Wright" <shane.wright@edigitalresearch.com> wrote:

>
> The RPMs installed from are these:
>
> postgresql-8.3.0-1PGDG.rhel4
> postgresql-libs-8.3.0-1PGDG.rhel4
> postgresql-server-8.3.0-1PGDG.rhel4

You need postgresql-debug (I think), Devrim?

Joshua D. Drake

>
> Thanks
>
> Shane


--
The PostgreSQL Company since 1997: http://www.commandprompt.com/
PostgreSQL Community Conference: http://www.postgresqlconference.org/
Donate to the PostgreSQL Project: http://www.postgresql.org/about/donate
PostgreSQL SPI Liaison | SPI Director |  PostgreSQL political pundit
Attachment

Re: Upgrade to 8.3.0?

From
Devrim GÜNDÜZ
Date:
Hi,

On Fri, 2008-02-29 at 21:31 +0000, Shane Wright wrote:
> I'm assuming that the default RPMs for RHEL 4 (on ftp.postgresql.org)
> are not built with --enable-cassert

Yeah, --enable-cassert is enabled only in beta packages.

Regards,
--
Devrim GÜNDÜZ , RHCE
PostgreSQL Replication, Consulting, Custom Development, 24x7 support
Managed Services, Shared and Dedicated Hosting
Co-Authors: plPHP, ODBCng - http://www.commandprompt.com/
Attachment