Re: RMAN Error
From: John Schaeffer <amerar_at_iwc.net>
Date: Wed, 25 Mar 2009 13:05:45 -0700 (PDT)
Message-ID: <4d6feedf-759f-4236-925e-920992bc5e79_at_k2g2000yql.googlegroups.com>
On Mar 25, 12:30 pm, jgar the jorrible <joel-ga..._at_home.com> wrote:
> On Mar 25, 10:00 am, johnbhur..._at_sbcglobal.net wrote:
>
> > On Mar 25, 12:34 pm, John Schaeffer <ame..._at_iwc.net> wrote:
>
> > snip
>
> > > > Personally I like to separate out delete expired functionality from
> > > > delete obsolete also ... ( in separate runs ).
>
> > > Not sure I quite understand the crosschecks before any backup
> > > commands, but I can try your suggestion.
>
> By doing this you are basically telling Oracle to be sure everything
> is where Oracle thinks it is, before it starts dinking around with
> things.
>
>
>
> > Cross checks before delete's is what I am recommending that you
> > consider. Maybe a factor in your original symptom maybe not.
>
> That's what I've wound up doing, since some stuff gets deleted outside
> of the automatic deletes, since the automatic deletes don't always
> work exactly as expected. Never quite figured out if it was a bug or
> something I do wrong, but am quite willing to assume it's me. I have
> several databases sharing a large device, and have had issues with not
> quite figuring the space allowance correctly (ie, the first time on a
> test db). Also what Oracle counts as a backup for purposes of
> redundancy can be confusing, show all rman settings and be sure they
> are understood.
>
> The state of the the packages ORA-4068 type of error can come from a
> number of things that cause recompilation, including changing grants,
> flushing shared pool, dependencies on other recompilation (which may
> not always be obvious), and a bunch of other things I can't think of
> right now. See metalink 778842.1 for some bugs. So it could be as
> simple as something getting aged out of the shared pool, then being
> called, you might consider enlarging shared pool some (just to push
> off shared pool fragmentation problems) and definitely should consider
> looking and seeing what is in there when.
>
> jg
> --
> _at_home.com is bogus.
> "Everyone makes mistakes sometimes."http://www3.signonsandiego.com/stories/2009/mar/25/1n25jet011211-pilo...
Date: Wed, 25 Mar 2009 13:05:45 -0700 (PDT)
Message-ID: <4d6feedf-759f-4236-925e-920992bc5e79_at_k2g2000yql.googlegroups.com>
On Mar 25, 12:30 pm, jgar the jorrible <joel-ga..._at_home.com> wrote:
> On Mar 25, 10:00 am, johnbhur..._at_sbcglobal.net wrote:
>
> > On Mar 25, 12:34 pm, John Schaeffer <ame..._at_iwc.net> wrote:
>
> > snip
>
> > > > Personally I like to separate out delete expired functionality from
> > > > delete obsolete also ... ( in separate runs ).
>
> > > Not sure I quite understand the crosschecks before any backup
> > > commands, but I can try your suggestion.
>
> By doing this you are basically telling Oracle to be sure everything
> is where Oracle thinks it is, before it starts dinking around with
> things.
>
>
>
> > Cross checks before delete's is what I am recommending that you
> > consider. Maybe a factor in your original symptom maybe not.
>
> That's what I've wound up doing, since some stuff gets deleted outside
> of the automatic deletes, since the automatic deletes don't always
> work exactly as expected. Never quite figured out if it was a bug or
> something I do wrong, but am quite willing to assume it's me. I have
> several databases sharing a large device, and have had issues with not
> quite figuring the space allowance correctly (ie, the first time on a
> test db). Also what Oracle counts as a backup for purposes of
> redundancy can be confusing, show all rman settings and be sure they
> are understood.
>
> The state of the the packages ORA-4068 type of error can come from a
> number of things that cause recompilation, including changing grants,
> flushing shared pool, dependencies on other recompilation (which may
> not always be obvious), and a bunch of other things I can't think of
> right now. See metalink 778842.1 for some bugs. So it could be as
> simple as something getting aged out of the shared pool, then being
> called, you might consider enlarging shared pool some (just to push
> off shared pool fragmentation problems) and definitely should consider
> looking and seeing what is in there when.
>
> jg
> --
> _at_home.com is bogus.
> "Everyone makes mistakes sometimes."http://www3.signonsandiego.com/stories/2009/mar/25/1n25jet011211-pilo...
I'll give it a try for tonight and see what happens. Received on Wed Mar 25 2009 - 15:05:45 CDT