| Oracle FAQ | Your Portal to the Oracle Knowledge Grid | |
 |
 | |||
| Oracle FAQ | Your Portal to the Oracle Knowledge Grid | |
|
| |||
Home -> Community -> Usenet -> c.d.o.server -> Re: Huge memory consumption during hot backup
Hi Howard,
I've configured RMAN, with a simple scripts like:
and the memory consumption during backup is about 30-40 Mb (about 1/15 than using the operating system scripts!).
Besides, I tried to use RMAN without a catalog, by using the command "rman target user/password_at_pbw nocatalog @d:\test_rman\script.txt log d:\backup\backup.log", and everything worked correctly. What do you mean exactly by "It's a distraction that requires management that gives you nothing very much in return"? I've read the "Oracle 8i - Recovery Manager User's Guide and Reference", and I've not seen any particular drawnback of using the control file instead of the catalog...
Best regards,
Paolo Borghesi
"Howard J. Rogers" <hjr_at_dizwell.com> wrote in message news:<414ac9dd$0$23895$afc38c87_at_news.optusnet.com.au>...
> DP wrote:
>
> > Paolo,
> >
> > RMAN does not require a second database (Repository) but it's a good
> > idea if at all possible.
>
> Disagree. It's a distraction that requires management that gives you nothing
> very much in return.
>
> > RMAN will store the recovery information in the control file, so you
> > will have to backup these files and protect them.
>
> Er, you would have to backup the control file(s) and protect it/them
> regardless of whether you are using RMAN or not.
>
> HJR
>
>
>
> > Good Luck,
> > Dennis
> >
> > Paolo wrote:
> >> Hi Howard,
> >> thank you very much for your reply. I've been posted the same
> >> message on Metalink and on my enterprise internal oracle support and I
> >> got no reply.
> >>
> >> I'll try to use RMAN, even if it is an invasive option (I've to create
> >> another database, isn't it?), and I'll let you know my results.
> >>
> >> Thank you again,
> >>
> >> Paolo Borghesi
> >>
> >>
> >> "Howard J. Rogers" <hjr_at_dizwell.com> wrote in message
> >> news:<414a29d3$0$23896$afc38c87_at_news.optusnet.com.au>...
> >>
> >>>Paolo wrote:
> >>>
> >>>
> >>>>Hi,
> >>>>I'm using Oracle 8.1.7.4.1 on Windows 2000 Advanced Server SP 3.
> >>>>The DB is in archive log mode, and I use the following script to
> >>>>perform an hot backup:
> >>>>
> >>>>-------------------------------------------------------
> >>>>spool temp_bk_res
> >>>>$SET backup=x:\backup
> >>>>connect user/password_at_mydb as sysdba
> >>>>$mkdir x:\backup\today
> >>>>alter tablespace GENESYS_CFG begin backup;
> >>>>$copy Y:\ORACLE\ORADATA\MYDB\USERS01.DBF x:\backup\today
> >>>>alter tablespace GENESYS_CFG end backup;
> >>>>alter tablespace GENESYS_LOG begin backup;
> >>>>$copy Y:\ORACLE\ORADATA\MYDB\USERS02.ORA x:\backup\today
> >>>>$copy Y:\ORACLE\ORADATA\MYDB\USERS02.ORA x:\backup\today
> >>>>$copy Y:\ORACLE\ORADATA\MYDB\USERS02.ORA x:\backup\today
> >>>>alter tablespace GENESYS_LOG end backup;
> >>>>[...]
> >>>>-------------------------------------------------------
> >>>>
> >>>>My problem is that when I begin the backup the available physical
> >>>>memory decreases by around 400-500Mb, and if my system have less free
> >>>>memory than requested it begins to swap a lot, making the database
> >>>>unable to reply to the application (which hang).
> >>>>
> >>>>I've tried also by replacing the "$copy" commands with the "$start
> >>>>/wait /low ocopy" commands, but the memory consumption has not
> >>>>changed; besides, I tried to change Windows' file cache usage, by
> >>>>setting the "File and Printer Sharing for Microsoft Networks"
> >>>>properties to "Maximize data throughput for network applications", but
> >>>>with no success.
> >>>>
> >>>>Do you have something to suggest?
> >>>>
> >>>>
> >>>>Best regards,
> >>>>
> >>>>Paolo Borghesi
> >>>
> >>>What I'd *really* suggest is to use RMAN to do your backups.
> >>>
> >>>But if RMAN is absolutely not on, then I'd suggest having a look at the
> >>>sort of Windows Scripting Host script I have available at
> >>>
> >>>http://www.dizwell.com/html/scripted_backups.html
> >>>
> >>>I've run that against many databases, and it doesn't have anything like
> >>>the memory hit you are talking about.
> >>>
> >>>(As the site says, I hope: although the script has one 9i specific bit in
> >>>it, it runs on 8i just fine so long as you edit the 21 lines bit to
> >>>whatever is appropriate to 8i).
> >>>
> >>>I'd be interested, in any case, to see whether you can replicate your
> >>>memory drop using it. If you can, something serious is up. If you can't
> >>>then it's something in the script you are already using.
> >>>
> >>>Regards
> >>>HJR
Received on Wed Sep 22 2004 - 05:41:39 CDT
 |
 |