Re: ok this sounds crazy BUT, is it a bug

From: Bradd Piontek <piontekdd_at_gmail.com>
Date: Tue, 4 Aug 2009 12:54:48 -0500
Message-ID: <e9569ef30908041054q499a70e9mc416efe75b1dfe66_at_mail.gmail.com>

I had researched the delay option a few weeks ago in relation to where to set it (log_archive_dest_n or on the recovery on the standby). The preferred method is on the primary (log_archive_dest_n). The standby recovery setting is left it for backward compatability, but deprecated I believe. I don't think using ARCH or LGWR implies Real-Time Apply. You can have LGWR (11g default, ARCH is being deprecated as well, I think) transport logs but not apply them. Real-Time apply is configured on the standby (alter database recover managed standby database using current logfile disconnect; , iirc). The Delay in log_archive_dest_n should allow for the redo stream to still be transmitted, but just delay when it gets applied.

Bradd Piontek
  "Next to doing a good job yourself,

        the greatest joy is in having someone
        else do a first-class job under your
        direction."

• William Feather

On Tue, Aug 4, 2009 at 12:45 PM, <TESTAJ3_at_nationwide.com> wrote:

>
> Mark, thanks, i'm thinking its a bug, its just interesting that I can say
> "use redo, not archive but do it on delay also",
>
> oh well just have to keep the issue in mind if/when we implement.
>
> thanks, joe
>
> _______________________________________
> Joe Testa, Oracle Certified Professional
> (Work) 614-677-1668
> (Cell) 614-312-6715
>
> Interested in helping out your marriage?
> Ask me about "Weekend to Remember"
> Dec 11-13, 2009 here in Columbus.
>
>
>
> From: "Mark W. Farnham" <mwf_at_rsiz.com> To: <TESTAJ3_at_nationwide.com>,
> "'Martin Klier'" <usn_at_usn-it.de> Cc: <oracle-l_at_freelists.org> Date: 08/04/2009
> 01:35 PM Subject: RE: ok this sounds crazy BUT, is it a bug
> ------------------------------
>
>
>
> So if you force Oracle into the Hobson’s choice of obeying your delay
> versus stopping updates on the primary they allow your delay to be
> comprimised to keep things rolling. I believe that is a bug. Here is my
> reasoning: Injection of a delay is a security and/or safety mechanism which
> must always trump performance (even a temporary hang.) Of course you can
> logically infinitely (and indefinitely within the space available) extend
> the wrap window by adding online log groups, so there is always and out.
> This is essentially the same situation as archiver stuck due to no available
> space on the archive device, and consistent logic should be applied in
> unwrapping the dilemna.
>
> So I think they should be making the other choice to the degree that I
> would call it a bug. On the other hand, even if you agree with me that it is
> a bug, it is a bug that for all practical purposes you can avoid.
>
> I hope what I just wrote makes sense to all y’all.
>
> Joe you push the coolest corner cases going.
>
> Regards,
>
> mwf
>
> ------------------------------
>
> *From:* oracle-l-bounce_at_freelists.org [
> mailto:oracle-l-bounce_at_freelists.org <oracle-l-bounce_at_freelists.org>] *On
> Behalf Of *TESTAJ3_at_nationwide.com*
> Sent:* Tuesday, August 04, 2009 12:23 PM*
> To:* Martin Klier*
> Cc:* oracle-l_at_freelists.org*
> Subject:* Re: ok this sounds crazy BUT, is it a bug
>
>
> Well i forced enough log switches to have gone thru all of the redo logs,
> and that forced it across, I'm thinking that even though I asked to redo log
> delay, its really falling back to archive logs apply, time to go and dig in
> the DRC*.logs.
>
> thanks, joe
>
> _______________________________________
> Joe Testa, Oracle Certified Professional
> (Work) 614-677-1668
> (Cell) 614-312-6715
>
> <snip>
>
>

--
http://www.freelists.org/webpage/oracle-l

Received on Tue Aug 04 2009 - 12:54:48 CDT