Обсуждение: GiST intarray Memory Allocation

Hi All



Platform: Win32 (XP SP2)

PG: 8.0.3 (default config file)



I have a table with circa 5.5Million rows and an int4[] column containing
about 28 elements each.



When I try to put a CREATE INDEX name ON table USING gist(column);



I get the following :



ERROR:  invalid memory alloc request size 2825271912



Any help appreciated or if you need more info just ask.



Thanks

Garrett

On Sun, Sep 04, 2005 at 04:49:27PM +0100, Garrett Heaver wrote:

> I have a table with circa 5.5Million rows and an int4[] column containing
> about 28 elements each.
>
> When I try to put a CREATE INDEX name ON table USING gist(column);
> I get the following :
>
> ERROR:  invalid memory alloc request size 2825271912

Hmm.  Have you tried a plain
SELECT * FROM table?

It sounds like your table may be corrupt.

Another alternative is that the opclass functions are buggy.  If the
table is not corrupt, you could try giving us complete working
instructions to try to reproduce the bug.

--
Alvaro Herrera -- Valdivia, Chile         Architect, www.EnterpriseDB.com
"Thou shalt not follow the NULL pointer, for chaos and madness await
thee at its end." (2nd Commandment for C programmers)

Performed full table select - worked perfectly

I can give you a dump of the table if that will help

Thanks
Garrett

-----Original Message-----
From: Alvaro Herrera [mailto:alvherre@alvh.no-ip.org]
Sent: 04 September 2005 17:09
To: Garrett Heaver
Cc: pgsql-bugs@postgresql.org
Subject: Re: [BUGS] GiST intarray Memory Allocation

On Sun, Sep 04, 2005 at 04:49:27PM +0100, Garrett Heaver wrote:

> I have a table with circa 5.5Million rows and an int4[] column containing
> about 28 elements each.
>
> When I try to put a CREATE INDEX name ON table USING gist(column);
> I get the following :
>
> ERROR:  invalid memory alloc request size 2825271912

Hmm.  Have you tried a plain
SELECT * FROM table?

It sounds like your table may be corrupt.

Another alternative is that the opclass functions are buggy.  If the
table is not corrupt, you could try giving us complete working
instructions to try to reproduce the bug.

--
Alvaro Herrera -- Valdivia, Chile         Architect, www.EnterpriseDB.com
"Thou shalt not follow the NULL pointer, for chaos and madness await
thee at its end." (2nd Commandment for C programmers)

Just to confirm also

I dumped the database and restored to a Suse Linux 9.3 Box

Same problem occurs (different alloc request size though)

I can forward a dump of the database if required

Cheers
Garrett

-----Original Message-----
From: pgsql-bugs-owner@postgresql.org
[mailto:pgsql-bugs-owner@postgresql.org] On Behalf Of Garrett Heaver
Sent: 04 September 2005 17:14
To: 'Alvaro Herrera'
Cc: pgsql-bugs@postgresql.org
Subject: Re: [BUGS] GiST intarray Memory Allocation

Performed full table select - worked perfectly

I can give you a dump of the table if that will help

Thanks
Garrett

-----Original Message-----
From: Alvaro Herrera [mailto:alvherre@alvh.no-ip.org]
Sent: 04 September 2005 17:09
To: Garrett Heaver
Cc: pgsql-bugs@postgresql.org
Subject: Re: [BUGS] GiST intarray Memory Allocation

On Sun, Sep 04, 2005 at 04:49:27PM +0100, Garrett Heaver wrote:

> I have a table with circa 5.5Million rows and an int4[] column containing
> about 28 elements each.
>
> When I try to put a CREATE INDEX name ON table USING gist(column);
> I get the following :
>
> ERROR:  invalid memory alloc request size 2825271912

Hmm.  Have you tried a plain
SELECT * FROM table?

It sounds like your table may be corrupt.

Another alternative is that the opclass functions are buggy.  If the
table is not corrupt, you could try giving us complete working
instructions to try to reproduce the bug.

--
Alvaro Herrera -- Valdivia, Chile         Architect, www.EnterpriseDB.com
"Thou shalt not follow the NULL pointer, for chaos and madness await
thee at its end." (2nd Commandment for C programmers)



---------------------------(end of broadcast)---------------------------
TIP 9: In versions below 8.0, the planner will ignore your desire to
       choose an index scan if your joining column's datatypes do not
       match

"Garrett Heaver" <garrett.heaver@researchandmarkets.com> writes:
> Just to confirm also
> I dumped the database and restored to a Suse Linux 9.3 Box
> Same problem occurs (different alloc request size though)
> I can forward a dump of the database if required

Please (unless you've already heard from Oleg or Teodor, who are
probably more qualified to debug it than I am).  Offlist of course.

            regards, tom lane