Re: A couple archive log / sequence number questions...
- From: "David Barbour" <david.barbour1@xxxxxxxxx>
- To: mason@xxxxxxxxxxx
- Date: Thu, 18 Oct 2007 13:55:11 -0400
Check the managed recovery trace file. It'll be in your bdump directory and
look something like <SID>_mrp0_<PID>.trc
On 10/17/07, Mason Loring Bliss <mason@xxxxxxxxxxx> wrote:
>
> Hi, all!
>
> I have a couple questions. I've started moving archive logs from our main
> database server to a number of standby servers. I noted that rsync on
> SuSE/
> SLES 9 (default operation / no checksumming) didn't notice a change in an
> archive log, resulting in that stand-by server getting out of date.
> Turning
> on checksumming for rsync, despite being harder on the disks, seems to
> have
> remedied that, but I'm wondering if there's a way to determine what
> archive
> logs have been applied.
>
> Two things I've found so far confuse me a bit.
>
> First, it seems like I can say this from sqlplus:
>
> select SEQUENCE#,APPLIED,STATUS from v$archived_log order by
> SEQUENCE#;
>
> On our primary (open, running) server, I see, at the end:
>
> SEQUENCE# APP S
> ---------- --- -
> 33632 NO A
> 33633 NO A
> 33634 NO A
> 33635 NO A
>
> On a stand-by, I see:
>
> ...
> 33632 NO D
> 33632 NO A
> 33633 NO D
>
> SEQUENCE# APP S
> ---------- --- -
> 33633 NO A
> 33634 NO D
> 33634 NO A
>
> I'm not sure why the "applied" column in "no" in each case. When I run
> through "recover database" in rman, it shows logs applying up until the
> end
> of the newly available data, and bringing the database up to test shows
> the
> data as being fresh, as expected. I'm also not sure why I'm seeing each
> sequence number twice on the stand-bys.
>
> I've also found that I can say "list archivelog all" in rman, but is that
> showing me what's on-disk, rather than what has been applied?
>
> Finally, when I say "recover database" it will give me an error saying
> that
> it's looking for a particular sequence number, after applying everything
> it
> can that's new, but running that just to elicit the error so I can see
> where
> we are seems a bit harsh. Plus, I can't do that on an open database, and
> I'd
> like to have one method to apply to both open *and* stand-by databases.
>
> What's the preferred method to query the sequence number of the last
> applied
> archive log on a stand-by database, so I can compare with the primary and
> with other stand-by databases to make sure no one is falling behind?
>
> Thanks in advance for clues!
>
> --
> Mason Loring Bliss (( If I have not seen as far as others, it is
> because
> mason@xxxxxxxxxxx )) giants were standing on my shoulders. - Hal
> Abelson
> --
> //www.freelists.org/webpage/oracle-l
>
>
>
Other related posts:
