Problem with lost archive logs

Thread view:

Enable EMail Alerts

Start New Thread

Thread rating:

tsteege@gmail.com - 30 Aug 2007 20:34 GMT

Not sure if this is the correct area to ask this question but it is a
start.

First of all I am new to Oracle so forgive me for how basic I will be.

We are running 9i on a 2003 server. Hard drive space is limited, and
of course filled up with archive logs. The all but one days worth of
logs were deleted using windows explorer. I found this out when my
Backup EXEC 11.d job crashed the following night.

Now I don't have any idea on how to tell Oracle/backkupexec that they
are gone and to stop looking for them.

Thanks in advance for any help.

Terry

Reply to this Message

tsteege@gmail.com - 30 Aug 2007 20:41 GMT

On Aug 30, 1:34 pm, "tste...@gmail.com" <tste...@gmail.com> wrote:

> Not sure if this is the correct area to ask this question but it is a
> start.
[quoted text clipped - 12 lines]
>
> Terry

Sorry I forgot to mention that the servers backup had not been running
for a while, and I can not recover the logs. Can I force Oracle to
think a backup has been done and then start my backup running nightly?

Thanks again.

Reply to this Message

Brian Peasland - 30 Aug 2007 20:55 GMT

> On Aug 30, 1:34 pm, "tste...@gmail.com" <tste...@gmail.com> wrote:
>> Not sure if this is the correct area to ask this question but it is a
[quoted text clipped - 19 lines]
>
> Thanks again.

How are you backing up your Oracle database? Are you just using Backup
Exec? If so, then your backups are inconsistent and may not be usable
for recovery. Have you tested recovery from these backups before?
Backups are useless unless you can recover using them.

I would recommend implementing RMAN to do your backup to disk. Then have
Backup Exec backup up the backup.

When you do a full backup of the Oracle database, you no longer need the
archived redo logs generated before that backup. So you can remove them.
RMAN can also be used to backup the archived redo logs and to delete
them after they have been backed up.

HTH,
Brian

Signature

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

Brian Peasland
dba@nospam.peasland.net
http://www.peasland.net

Remove the "nospam." from the email address to email me.

"I can give it to you cheap, quick, and good.
Now pick two out of the three" - Unknown

--
Posted via a free Usenet account from http://www.teranews.com

Reply to this Message

joel garry - 30 Aug 2007 21:45 GMT

> tste...@gmail.com wrote:
> > On Aug 30, 1:34 pm, "tste...@gmail.com" <tste...@gmail.com> wrote:
[quoted text clipped - 33 lines]
> RMAN can also be used to backup the archived redo logs and to delete
> them after they have been backed up.

You know, this thing about not needing archived logs generated before
the backup isn't strictly true.

I have a situation where someone incorrectly added a datafile, then
put it offline. Weeks later, here I am trying to restore a standby
from a fresh complete RMAN backup, can't do, need the archive log from
the time that datafile was created/offlined, don't have. There's well-
known ways to open it if you use resetlogs, but that's inappropriate
for a 9206 standby. It's the only file the restored db thinks is
fuzzy.

If anybody knows a way out of this short of dropping the datafile and
redoing the backup/restore, my labor day weekend is all ears.

(And now I remember why I always prefer to keep all archive logs
redundantly forever, even with a much less severe SLA).

jg
--
@home.com is bogus.
"Wonder Bread is Toast"
"Market Stale for Wonder Bread" - newspaper headlines.

Reply to this Message

vitalisman@gmail.com - 31 Aug 2007 11:03 GMT

> If anybody knows a way out of this short of dropping the datafile and
> redoing the backup/restore, my labor day weekend is all ears.

AFAIK it is not possible.

Out of curiosity, were these archived logs deleted by RMAN? Which
command exactly?

Reply to this Message

joel garry - 31 Aug 2007 15:07 GMT

On Aug 31, 3:03 am, vitalis...@gmail.com wrote:

> > If anybody knows a way out of this short of dropping the datafile and
> > redoing the backup/restore, my labor day weekend is all ears.
[quoted text clipped - 3 lines]
> Out of curiosity, were these archived logs deleted by RMAN? Which
> command exactly?

No, they are deleted outside of RMAN. Each RMAN backup has at least
the last 3 days worth of logs, but not every backup is kept. Not my
recommended way of doing things, but it took major hardware failures
of the standby storage plus several manual mistakes (while I was off
galavanting through the giant redwoods) to get to this situation.
It's not that bad of a situation since the primary production system
only has a minor interruption coming because I have to drop and
recreate the tablespace with the bogus offline file. And I charge by
the half-hour :-) (I did make some mistakes too, one of which was not
dealing with this bogus file as soon as I became aware of it - it
didn't seem to be causing any issues :-O ).

I'm not entirely sure, but I think the newer versions of RMAN being
able to restore across incarnations might have a way around this sort
of problem. That doesn't help me this weekend.

jg
--
@home.com is bogus.
I want my flying car! http://news.bbc.co.uk/1/hi/business/6970031.stm

Reply to this Message

vitalisman@gmail.com - 31 Aug 2007 15:43 GMT

> On Aug 31, 3:03 am, vitalis...@gmail.com wrote:
>
[quoted text clipped - 8 lines]
> No, they are deleted outside of RMAN. Each RMAN backup has at least
> the last 3 days worth of logs, but not every backup is kept.

OK. I feared RMAN might have "unfortunately" deleted the required
archived log backups during the execution of 'delete obsolete' in your
scenario.

Does it complain about your datafile when you run "report
unrecoverable;"?

> I'm not entirely sure, but I think the newer versions of RMAN being
> able to restore across incarnations might have a way around this sort
> of problem. That doesn't help me this weekend.

Which incarnations? I can't see any resetlogs in your scenario.

Jerome

Reply to this Message

joel garry - 31 Aug 2007 19:17 GMT

On Aug 31, 7:43 am, vitalis...@gmail.com wrote:

> > On Aug 31, 3:03 am, vitalis...@gmail.com wrote:
>
[quoted text clipped - 15 lines]
> Does it complain about your datafile when you run "report
> unrecoverable;"?

No, I would guess because it is offline.

> > I'm not entirely sure, but I think the newer versions of RMAN being
> > able to restore across incarnations might have a way around this sort
> > of problem. That doesn't help me this weekend.
>
> Which incarnations? I can't see any resetlogs in your scenario.

The idea is to avoid resetlogs to keep the standby going. The normal
answer to not being able to recover a db with this sort of file
problem includes open resetlogs. There are lots of things that can't
be done with normal advice if a standby is involved. Anything that
avoids generating redo for performance and anything requiring a
resetlogs are two major ones.

> Jerome

jg
--
@home.com is bogus.
What's in a name? http://www.guardian.co.uk/worldlatest/story/0,,-6865184,00.html

Reply to this Message

Jerome Vitalis - 31 Aug 2007 19:34 GMT

> On Aug 31, 7:43 am, vitalis...@gmail.com wrote:
>> OK. I feared RMAN might have "unfortunately" deleted the required
[quoted text clipped - 5 lines]
>
> No, I would guess because it is offline.

Hmm, I don't think so.
But I've just re-read the reference guide for "report unrecoverable" and
it seems to take unrecoverable operations (when no redo is generated)
into account only. So that would explain the output of the command in
your case.

> The idea is to avoid resetlogs to keep the standby going. The normal
> answer to not being able to recover a db with this sort of file problem
> includes open resetlogs. There are lots of things that can't be done
> with normal advice if a standby is involved. Anything that avoids
> generating redo for performance and anything requiring a resetlogs are
> two major ones.

Thanks Joel for clarifying that point.

Reply to this Message