Re: Archived Log File broken

FlameDance <FlameDance_at_gmx.de> wrote in message news:<cbtphv$aci$01$1_at_news.t-online.com>...
> Sybrand Bakker schrieb:
> >
> >
> >>What can be the cause of this error?
> >
> > Hacking your own 'standby' mechanism together, where a stable Oracle
> > solution exists.
>
> I believe I did it by the book. Which would be your suggested stable
> Oracle solution? A connection between production and standby via
> SQLNet8? That would mean instant recovery, wouldn't it? It's not what I
> want.
>
> > I guess the directory is NFS mounted.
>
> Yes.
>
> > Oracle has always warned *against* using NFS,
>
> The solution Oracle advises is to use a directory on the production
> machine and (if available) hardware or OS mirroring on that machine.
> Neither solution helps when the production machine burns down.
>
> > in the documentation no one cares to read.
>
> Mr. Sybrand, I waded through about 1000 pages of documentation for this.
> Documentation which is not always clear, documentation parts which at
> times contradict each other.
>
> Maybe I missed something? I wouldn't bet that I didn't - but I don't
> care for your bullying attempts.
>
> Stephan

If you are going to continue to try your solution, you need to add a mechanism to compare the production and standby archived logs before applying to standby. As you've seen, NFS does not guarantee that every bit will make it over, and can and will corrupt these critical data streams. Check if your NFS uses UDP, and look up what UDP means.  And even TCP based NFS can screw up.

jg

--
@home.com is bogus.
I posted a more complete answer to this yesterday, but that set off
the google spam alarm, which mails the post to my non-extant spam
attractor address rather than posting.  How ironic.