The other parameters that will get you in the same trouble are related to the special buffer
pools.. I ran into this issue with 9iR1 and decided to stay away from SPFILE stuff...
The failure of Oracle (programmers) checking for such exclusiveness (?) tells their background.
Doesn't it?
I forget where I read this (may a post right here), but, someone suggested to create a database
trigger (to fire before shutdown) to write out pfile from spfile. Good idea, I think. And keep a
backup of the init.ora.
SGA_MAX_SIZE is yet another such blunder. It is supposed to set the high water mark for the SGA
size, so one can "dynamically" change memory allocation to certain SGA components.
On SUN Solaris (with ISM) it's useless. And, just today I found out that it does not work with AIX
5L either. Later tonight I will test it with AIX 4.3. The instance started with SGA sized to the
value specified by this parameter.
- Kirti
- Arup Nanda <orarup_at_hotmail.com> wrote:
> This can be solved by creating the PFILE again from the SPFILE; and this
> doesn't need the instance to be up. However, you did drive the nail home,
> Niall, with this example.
>
> Part of the problem could be avoided by restricting the ALTER SYSTEM, SYSDBA
> or SYSOPER privileges; but in a large shop with a lot of production support
> DBAS....!!
>
> Arup Nanda
>
> ----- Original Message -----
> To: "Multiple recipients of list ORACLE-L" <ORACLE-L_at_fatcity.com>
> Sent: Wednesday, July 16, 2003 4:59 PM
>
>
> > Not only that but this sort of thing should be avoidable.
> >
> > SQL> alter system set db_cache_size=10m scope=both;
> >
> > System altered.
> >
> > <time passes>
> >
> > SQL> alter system set db_block_buffers=1000 scope=spfile;
> >
> > System altered.
> >
> > SQL> shutdown;
> > Database closed.
> > Database dismounted.
> > ORACLE instance shut down.
> > SQL> startup
> > ORA-00381: cannot use both new and old parameters for buffer cache size
> > specific
> > ation
> > SQL>
> >
> >
> > What do you do if you didn't have an old fashioned text file? I'd also
> > like to know how spfile changes get documented, it scares me enough that
> > someone can change parameters on the fly, let alone that that can be
> > done without comment or recording who did it. I'll admit thugh that this
> > last is actually a failure of management and not technology.
> >
> > Niall
> > > -----Original Message-----
> > > From: ml-errors_at_fatcity.com [mailto:ml-errors_at_fatcity.com] On
> > > Behalf Of Arup Nanda
> > > Sent: 16 July 2003 06:09
> > > To: Multiple recipients of list ORACLE-L
> > > Subject: Re: How to make SPFILE in sync with INIT.ORA ?
> > >
> > >
> > > Dennis,
> > >
> > > > My understanding is that Oracle felt if it was going to be a
> > > > 24x7x365 database, it had to have more parameters that could be
> > > > changed online
> > >
> > > I couldn't agree more. 9i had more parameters that can be
> > > changed online; hope there are more in line.
> > >
> > > > it doesn't make sense to rely on the DBA
> > > > to remember to also update the init.ora file. You should be able to
> > > > just make the change once.
> > >
> > > Absolutely. That's the point I was trying to raise. Oracle
> > > did provide the spfile route to make _persistent_ changes on
> > > the fly, but for only those parameters that can be changed
> > > through alter system set ... The problem was there is a
> > > different file that needs to be updated when making those
> > > other parameter changes. There should be only one place to
> > > change - hopefully spfile - either through the alter system
> > > command or othe editing. This editing could be done through a
> > > GUI interface, too, if needed; although, call me
> > > old-fashioned (and I am old, anyway, at least in IT
> > > timeline), personally I prefer the vi editor. And this is not
> > > an impossible idea. I gave the example of the listener.ora
> > > file. It can be edited (so old fashioned), through the
> > > lsntctl SAVE_CONFIG command (so spfile-like) or through Net
> > > Assistant (so GUI-friendly). The end result is the same - one
> > > file - regardless of how you modify a parameter. The same
> > > approach could have been done in pfile-spfile case.
> > >
> > > > I think with the pfile and spfile, Oracle was trying to
> > > give us the
> > > > best of both worlds.
> > >
> > > Hmm! You think so? I think they just gave us a half-cooked
> > > hair-brained split-pea honey-mustard-ketchup-coated germ of a
> > > solution - sort of like Oracle 6 _without_ the Transaction
> > > Processing Option, if anyone remembers that. After you wipe
> > > off all those condiments, you find a Dr. Jekyll & Mr. Hyde
> > > parameter file(s)!
> > >
> > > Regards,
> > >
> > > Arup
> > >
> > >
> > > ----- Original Message -----
> > > To: "Multiple recipients of list ORACLE-L" <ORACLE-L_at_fatcity.com>
> > > Sent: Wednesday, July 16, 2003 12:24 AM
> > >
> > >
> > > > Arup
> > > > My understanding is that Oracle felt if it was going to be a
> > > > 24x7x365 database, it had to have more parameters that could be
> > > > changed online.
> > > Some
> > > > parameters were made changeable in Oracle9i, with hints of
> > > more to come.
> > > > If the future is online changes, it doesn't make sense
> > > to rely on
> > > > the
> > > DBA
> > > > to remember to also update the init.ora file. You should be able to
> > > > just make the change once.
> > > > Also, there is a new generation of I.S. people coming along that
> > > doesn't
> > > > think you can operate a computer without a mouse. Against
> > > Microsoft's
> > > super
> > > > GUI interfaces, the idea of manually editing a text
> > > configuration file
> > > seems
> > > > very last century.
> > > > I think with the pfile and spfile, Oracle was trying to
> > > give us the
> > > best
> > > > of both worlds. How well they succeeded is a matter of judgement.
> > > >
> > > > Dennis Williams
> > > > DBA
> > > > Lifetouch, Inc.
> > > > dwilliams_at_lifetouch.com
> > > >
> > > > -----Original Message-----
> > > > Sent: Tuesday, July 15, 2003 9:14 PM
> > > > To: Multiple recipients of list ORACLE-L
> > > >
> > > >
> > > > The ability to change the system parameters without bouncing the
> > > > system is not provided by spfile, neither in RAC nor single
> > > instance
> > > > databas. It depends upon the the parameter that can be changed
> > > > dynamically or not.
> > > >
> > > > I guess you wanted to convey the impression that the using
> > > spfiles the
> > > > parameters can be changed and the changes can be persistent across
> > > > shutdowns. But that is not just in RAC; it's true for
> > > single instance
> > > > DBs, too.
> > > >
> > > > Now, suppose you want to set a parameter that can't be
> > > changed using
> > > > ALTER SYSTEM, such as, say, java_pool_size. How do you plan to make
> > > > the change? You have to open up the old favorite init.ora file and
> > > > start the database with pfile=init.ora option. At that stage the
> > > > spfile is not active and
> > > your
> > > > issuing Alter system set db_cache_size = 800m scope = [ memory |
> > > > spfile | both ] sid = * has no effect. You must create the
> > > spfile from
> > > > the pfile
> > > and
> > > > then use the newly created spfile to use this "dynamic" parameter
> > > > persistent. Note the complexity involved - spfile allowed
> > > you to make
> > > > the changes to some parmeters using alter system
> > > persistent; but for
> > > > all other parametrs you are forced to use pfile. What
> > > happens if you
> > > > ommit the pfile=init.ora clause? The database will pickup
> > > the spfile,
> > > > which will
> > > _not_
> > > > have your changes.
> > > >
> > > > To fully appreciate the value of the spfile parameter,
> > > Oracle should
> > > > have allowed editing spfile directly and completely done away with
> > > > pfile. Splitting functionality across two different implementations
> > > > adds to difficulties, does not resolve them. I hope future
> > > versions of
> > > > Oracle do offer that functionality. It's not that difficult
> > > or unheard
> > > > of, either. Listener.ora, tnsnames.ora are all editable and
> > > also read
> > > > by Net
> > > Assistant.
> > > >
> > > > just my .02
> > > >
> > > > Arup Nanda
> > > >
> > > >
> > > > ----- Original Message -----
> > > > To: "Multiple recipients of list ORACLE-L" <ORACLE-L_at_fatcity.com>
Received on Wed Jul 16 2003 - 17:12:40 CDT
