<long response>
My opinion that shutdown abort is totally safe is
based on this:
The only thing that matters in the database is redo
logs. Everything else is "optional" - ie it could be
recovered albeit with difficulty (eg you can recreate
a controlfile from scratch if you really have to, and
datafiles can be recovered as long as you have all of
the required redo stream). In fact, datafiles
themselves can be thought of as simply "performance
enhancement" - ie you could (theoretically) run all
your queries by trawling through all the redo that has
even been recorded in the db; the datafiles simply are
an "aid" to improve the performance of this.
And redo log operations are synchronous and "atomic"
from Oracle's perspective - something is either in
them or its not, there is no "maybe". shutdown abort
has no bearing on the synchronicity of redo operations
and hence is totally safe...
...BUT...does this mean 'abort' mode should be for
every shutdown - well this depends on your
circumstances. Most cluster software will deem
'shutdown abort' as catastrophic and invoke all sorts
of failover routines - so maybe abort is not the way
to go in these cases. Similarly, the cold backups
around a db upgrade should not be done via shutdown
abort, because those redo logs (which are so critical)
might undergo a structure change.
So my 'mandate' (for lack of a better term) is for
those rare occasions when you actually need to
shutdown a database, you should issue a checkpoint,
then a shutdown abort
hth
connor
- kkennedy <kkennedy_at_firstpoint.com> wrote: > I
have steel belted radial tires on my car that are
> supposed to be puncture resistant. Is this a good
> reason for me to go out of my way to drive by a
> construction site every morning? By my way of
> thinking, no. If my regular road is blocked and I
> have no alternative, then I will drive by the
> construction site reasonably confident that the
> debris will not puncture my tires. If I'm in a big
> hurry and driving by the construction site is
> significantly quicker, then I will consider it.
> But, I don't go out of my way looking for trouble.
>
> Does anyone have a better argument than "I've been
> doing this for years and it has always worked?"
> Kevin Kennedy
> First Point Energy Corporation
>
> -----Original Message-----
> Sent: Tuesday, July 23, 2002 4:04 PM
> To: Multiple recipients of list ORACLE-L
> imm
>
>
> I would have to agree with Rich ....... For years I
> have been using startup force restrict, shutdown
> normal prior to cold backups .... with the
> assumption that I want a completely 'static' and
> normal shutdown prior to the backup. Because of
> applications that maintain persistent connections,
> daemons that re-connect, etc. etc. that do not 'let
> loose' of the database ..... I have on numerous
> occasions had shutdown immediate 'hang'. I believe
> the operative process is the startup force restrict
> that enables auto-recovery and then the normal
> shutdown. I have never had an subsequent problems.
> Why chance any anomalous conditions that could be
> introduced to the recovery process or rebuild of a
> database.
> -----Original Message-----
> Rich
> Sent: Tuesday, July 23, 2002 1:21 PM
> To: Multiple recipients of list ORACLE-L
> imm
>
>
> You are assuming that shutdown immediate is a fix
> amount of time. If there is a lot to rollback it
> may take longer to rollback (via shutdown immediate)
> than the parallel crash recovery (the startup after
> a shutdown abort).
> It depends on the system. The question was why is
> immediate better than abort? I wanted to challenge
> the assumption that it is not safe to do an abort.
> You are relying on Oracle's recovery mechanisms
> which I have to assume are reliable. It is after
> all one of the reasons I prefer Oracle.
> - Rich
>
> -----Original Message-----
> Sent: Tuesday, July 23, 2002 3:34 PM
> To: Multiple recipients of list ORACLE-L
> imm
>
>
> Well.....
>
> I'll agree with you only on the basis that shutdown
> immediate sometimes hangs and in those cases it is
> quicker to do the abort/start/shut normal
> combination. However, based on a quick review of my
> logs from last night (cold backup), I see the
> shutdown immediate took about 12 seconds. The
> following startup (which needed no recovery) took
> about a minute.
>
> Had I used the shut abort technique, I expect I
> would have seen, let's say 5 seconds for the shut
> abort, 60 seconds or so for the startup restricted,
> then about 12 seconds for the shutdown normal.
>
> Hmmm. Doesn't seem so cut and dried to me. I think
> I'll keep using my shutdown script that tries
> shutdown immediate and only does the abort, etc. if
> immediate takes too long. At this site, the
> shutdown immediate only seems to fail about once a
> month. I can live with that unless someone comes up
> with a more compelling reason why the shutdown abort
> is better than a shutdown immediate. So far, all
> I've seen is the argument that shutdown abort is not
> evil -- I'm not one who thinks it is evil, I'm just
> not convinced that it is somehow better.
> Kevin Kennedy
> First Point Energy Corporation
>
> -----Original Message-----
> Sent: Tuesday, July 23, 2002 11:14 AM
> To: Multiple recipients of list ORACLE-L
> imm
>
>
> I don't necessarily agree that shutdown immediate is
> quicker. If you force a checkpoint prior to the
> shutdown abort the subsequent crash recovery upon
> startup is usually pretty fast. Parallel recovery
> could be a factor as well.
> - Rich
>
> -----Original Message-----
> Sent: Tuesday, July 23, 2002 1:14 PM
> To: Multiple recipients of list ORACLE-L
> immedia
>
>
> For openers, shutdown immediate is generally quicker
> than the combination of
> shutdown-abort/startup-restrict/shutdown-normal. It
> is also gentler. Consider the analogy of shutting
> down a Windows desktop computer. Is it preferable
> to do a standard software shutdown (and maybe tell
> Windows that you really want to end that hung
> process) or is it preferable to yank the plug out of
> the wall then plug it back in again, start up the
> machine, then shut it down gracefully? I always try
> to shut Windows down gracefully and only pull the
> plug when the damn thing is too stupid or brain dead
> to figure out what shutdown means. I do the same
> with Oracle.
> Kevin Kennedy
> First Point Energy Corporation
>
> -----Original Message-----
> Sent: Monday, July 22, 2002 7:53 PM
> To: Multiple recipients of list ORACLE-L
> immedia
>
>
>
> > -----Original Message-----
> > From: April Wells [ mailto:awells_at_csedge.com]
> >
> > The solutions (the ones that I got) aren't good
> ones.
> >
> > Shutdown abort/startup restricted/ shutdown
> immediate... (a 'VALID
> > solution'???)
>
> This might be a naive question, but why is
> -> shutdown immediate
> better than
> -> shutdown abort / startup restrict / shutdown
> normal ?
>
> (That is assuming of course that no user / job will
> try to sneak in after you do the startup restrict)
>
>
Connor McDonald
http://www.oracledba.co.uk
http://www.oaktable.net
"Remember amateurs built the ark - Professionals built the Titanic"
Do You Yahoo!?
Everything you'll ever need on one web page
from News and Sport to Email and Music Charts
http://uk.my.yahoo.com
--
Please see the official ORACLE-L FAQ: http://www.orafaq.com
--
Author: =?iso-8859-1?q?Connor=20McDonald?=
INET: hamcdc_at_yahoo.co.uk
Fat City Network Services -- (858) 538-5051 FAX: (858) 538-5051
San Diego, California -- Public Internet access / Mailing Lists
--------------------------------------------------------------------
To REMOVE yourself from this mailing list, send an E-Mail message
to: ListGuru_at_fatcity.com (note EXACT spelling of 'ListGuru') and in
the message BODY, include a line containing: UNSUB ORACLE-L
(or the name of mailing list you want to be removed from). You may
also send the HELP command for other information (like subscribing).
Received on Wed Jul 24 2002 - 04:48:21 CDT
