Обсуждение: PostgreSQL VACCUM killing CPU

Hi all !

I have got such problem.
Im running Postgresql 7.3.2 on Linux 2.6.13.
What is see when VACCUM is running and killing my CPU is:

Cpu(s):  3.2% us,  0.0% sy,  0.0% ni,  0.0% id, 96.8% wa,  0.0% hi,
0.0% si

what i am worry about is "96.8% wa" why is it like that?

what is the process waiting for ?

could somone explain me that please? :)

Best regards
david

The "wa" means waiting on IO.  Vacuum is a very IO intensive
process.  You can use tools like vmstat and iostat to see how much
disk IO is occurring.  Also, sar is very helpful for trending these
values over time.

-- Will Reese   http://blog.rezra.com
On May 9, 2006, at 5:19 AM, blender@jaskom.pl wrote:

> Hi all !
>
> I have got such problem.
> Im running Postgresql 7.3.2 on Linux 2.6.13.
> What is see when VACCUM is running and killing my CPU is:
>
> Cpu(s):  3.2% us,  0.0% sy,  0.0% ni,  0.0% id, 96.8% wa,  0.0% hi,
> 0.0% si
>
> what i am worry about is "96.8% wa" why is it like that?
>
> what is the process waiting for ?
>
> could somone explain me that please? :)
>
> Best regards
> david
>
>
> ---------------------------(end of
> broadcast)---------------------------
> TIP 5: don't forget to increase your free space map settings

>> I have got such problem.
>> Im running Postgresql 7.3.2 on Linux 2.6.13.

Also, you should seriously consider upgrading. 8.1.3 is the current
PostgreSQL release. If you must remain on 7.3, at least upgrade to
7.3.14, which contains *many* bugfixes.

Michael Glaesemann
grzm seespotcode net

>> I have got such problem.
>> Im running Postgresql 7.3.2 on Linux 2.6.13.

Also, you should seriously consider upgrading. 8.1.3 is the current
PostgreSQL release. If you must remain on 7.3, at least upgrade to
7.3.14, which contains many bugfixes.

Michael Glaesemann
grzm seespotcode net

On Tue, May 09, 2006 at 03:19:08AM -0700, blender@jaskom.pl wrote:
> I have got such problem.
> Im running Postgresql 7.3.2 on Linux 2.6.13.
> What is see when VACCUM is running and killing my CPU is:
>
> Cpu(s):  3.2% us,  0.0% sy,  0.0% ni,  0.0% id, 96.8% wa,  0.0% hi,
> 0.0% si
>
> what i am worry about is "96.8% wa" why is it like that?

It's killing your disk drives instead of CPU(which is mostly _idle_
waiting for I/O completion).

Run this command to get an idea of the I/O activities:
        iostat -x 3 3

[AD]Running a kernel patched with adaptive read-ahead may help it:
http://www.vanheusden.com/ara/adaptive-readahead-11.1-2.6.16.5.patch.gz