Re: How to housekeep archive log files?

Thanks Howard for the invaluable experience.

I'm also looking if anyone can share a (or portion of) Unix script to move non-current archive files to another bucket. (Noting that the current archive file has to stay for the db to use) It would be a great help if anyone hv a sample.

thanks,
jackling.

Howard J. Rogers ¼¶¼g©ó¤å³¹ <3ab9bdca$1_at_news.iprimus.com.au>...
>
>"PasirRis" <jackling_at_singnet.com.sg> wrote in message
>news:99advh$mnp$1_at_violet.singnet.com.sg...
>> version: Oracle Std Ed 8.1.7 on Sun Solaris 8.
>>
>>
>> 'archive log list' tells me some archive log seq nos. e.g. from 132 and
>> current is 149.
>> I do a hot backup of datafiles
>> then, I issue the command 'alter system log switch'. (so, the current
 arch.
>> log file will be 150 and files up to 149 is the targeted group for
 backup)
 I
>> backup all the archive log files in the directory.
>>
>> I suppose after the datafiles, archive log files are backup (say to
 tape),
 I
>> don't need to keep them in the disk and can remove those 'old' archive
 log
>> files already.
>>
>> Question:
>>
>> (Any sample?) How to write a script to remove those archive log files
>> already backuped? In the sample above, I suppose I should keep archive
 files
>> from 150 onwards on the disk but anything b4 it can be removed from disk
>> SAFELY (?).
>>
>> Any good idea?
>
>
>Here's a good idea: buy more hard disks, and don't be too keen to get rid
 of
>prior archives. Yes, technically, you can safely remove any archives
>created before your lastest hot backup cycle started, but that presupposes
>that you will always be able to restore from the latest backup. What
>happens if you discover the latest backup is corrupt, or unuseable, or is
>inadvertently wiped by an eager Junior DBA? What happens if a User comes
 to
>you and says "I'm sorry, I dropped the most important table in the system!"
>"When did you do it?" "Three days ago, but I only just noticed"?
>
>I tend to subscribe to the 'age out on disks and then onto tape' theory:
>Script a backup such that the current set of archives are copied into a
 DAY1
>bucket. The ones that were in DAY1 get transferred to the DAY2 bucket.
 The
>ones previously in the DAY2 bucket get copied into the DAY3 bucket, and so
>on, until the ones previously in the DAY5 bucket get archived onto tape.
>You then have the usual grandfather-father type of tape rotation that
>ensures recoverability of old archives for up to, say, 6 months.
>
>I actually used to do the same for the hot backups themselves, too.
>
>The more you can keep on disk, the quicker recoveries will be. No nasty
>serial access devices to deal with (and to slow the recovery process down).
>
>But as for archives: they are the only thing you have protecting you from
>data loss. I think they deserve a little cossetting as a result.
>
>Regards
>HJR
>
>
>
>
>>
>> thanks,
>> jackling.
>>
>>
>
>
Received on Thu Mar 22 2001 - 08:00:24 CST