Run out of QSF

[N SUMMERSALL]

I am running a complex update in v6/04 that crashes out and gives me the
message
QSF dynamic memory pool is exhausted (E_QS0001).

Is there anything I can do to increase the QSF?

Thanks

Neil

[Roy Hann]

N.Summ...@bradford.ac.uk (N SUMMERSALL) writes:
:
: I am running a complex update in v6/04 that crashes out and gives me the

: message
: QSF dynamic memory pool is exhausted (E_QS0001).
:
: Is there anything I can do to increase the QSF?

I am operating from memory (and my pool is nearly exhausted :-), but
I think there are two ways to fix this. The size of the QSF memory
pool can be set by you, or the server can estimate how big it should
be based on the number of connected sessions you tell it to
expect in $II_SYSTEM/ingres/files/rundbms.opt. Either bump up the
number of sessions in that file, or use the -qsf.pool_size option to
set the pool size, then shutdown and re-start the server.

Read the Installation and Operations guide.

========================================================================

Roy Hann b...@supernet.ab.ca
BKB Engineering Ltd., (MIME-capable mail agent)
11211-76th Avenue,
Edmonton, Alberta Tel: (403)438-2531 (office)
T6G 0K2 (403)437-0860 (home)
Canada FAX: (403)437-3367

========================================================================

[Vincent Catry]

/ hpgnd:comp.databases.ingres / N.Summ...@bradford.ac.uk (N SUMMERSALL) / 10:50 am Jun 13, 1994 /

Thanks

Neil
----------

Yes,

You can increase parameter qsf.pool_size
in $II_SYSTEM/ingres/files/rundbms.opt
ie:

-qsf.pool_size 3000000
(in bytes)

default is number of connected sessions * 40K + 60K

Vincent
(ca...@chamonix.grenoble.hp.com)

[Wal Brown]

...
: expect in $II_SYSTEM/ingres/files/rundbms.opt. Either bump up the

: number of sessions in that file, or use the -qsf.pool_size option to
: set the pool size, then shutdown and re-start the server.

You can monitor QSF pool usage with:

SET TRACE POINT QS501;

This is a 'pseudo-trace point': it doesn't actually set anything. Use it
in the TM and the current QSF pool size and %used are logged to the screen.

There is also a QS502, QS503, etc. The info gets more esoteric as they
go on.

Angela Lamb

[Gerrard Hocks]

In article <CrFpz...@oasis.icl.co.uk>, Wal Brown <w...@oasis.icl.co.uk> wrote:
>...
>: expect in $II_SYSTEM/ingres/files/rundbms.opt. Either bump up the
>: number of sessions in that file, or use the -qsf.pool_size option to
>: set the pool size, then shutdown and re-start the server.
>
>You can monitor QSF pool usage with:
>
> SET TRACE POINT QS501;
>

>Angela Lamb
>

Has anyone noticed that if you set this trace point with an Ingres user that
is NOT permitted to set trace points you lose a bit of your QSF memory pool ?
Each time you run it, you lose QSF memory. An Ingres shutdown seems to be
required to restore your installation..

Gerrard.

----------------------------------------------------------------------------
Gerrard Hocks ///\ * E-Mail ger...@comsense.com
Common Sense Computing ///\\\///\ * International +61 3 885 5829
16 Dunlop Street \\\///\\\/ * USA & Canada 1-800-588-8869
Ashburton VIC AUS 3147 \_\/ * United Kingdom 0-800-895-639
----------------------------------------------------------------------------

[Gavin Trigg]

>
>
> I am running a complex update in v6/04 that crashes out and gives me the
> message
> QSF dynamic memory pool is exhausted (E_QS0001).
>
> Is there anything I can do to increase the QSF?
>
> Thanks
>
> Neil
>

RTFM. The Installation & Operation Guide gives details on the dbms server
parameter -qsf.pool_size.

--
Gavin Trigg Internet: ga...@asis.unimelb.EDU.AU
Telephone: +61 3 344 6003 Fax +61 3 347 4803
Data Base Administrator, Information Technology Services (ASIS),
The University of Melbourne, Parkville, Australia. 3052.

[A division of ms13.mis.semi.harris.com]

Gavin Trigg (ga...@asis.unimelb.EDU.AU) wrote:
: >
: >
: > I am running a complex update in v6/04 that crashes out and gives me the

I've seen this problem before recently on one my machines. It was caused by
a user running a query that had a very large QEP. Since QEPs must be stored
contiguously in the QSP pool, and this machine always had several users running
nasty ad-hoc queries, increasing the server's qsf pool parm didn't always help.
I would suggest turning on the printqry trace point, disk space permitting, to
log whats running at what time. Then when the QSF fills up, look in the dbms
server's log file (errlog.log) to find the time when this happened.
Then you can match up the times to find which query is the culprit.

I was able to re-write the user's query, simplifying it.
--
Tom Gillespie _ _
Harris Semiconductor | | | |
Manufacturing Systems __/\/\/\
Palm Bay, FL | | | |
Tom.Gi...@harris.com _ _