Re: spreading my wings -- looking for some recommdations/best practices

From: Howard J. Rogers <hjr_at_dizwell.com>
Date: Fri, 22 Oct 2004 06:28:09 +1000
Message-Id: <41781bd0$0$23895$afc38c87@news.optusnet.com.au>

Noons wrote:

> "Howard J. Rogers" <hjr_at_dizwell.com> wrote in message
> news:<41771675$0$23894$afc38c87_at_news.optusnet.com.au>...
>
>

>> Again, I don't know what happens in Unix if you modify the link whilst
>> it's in use (I wonder if it's a bit like what happens when you delete the
>> control files from a currently-running instance.... and the database
>> keeps on chugging away perfectly normally).

>
> I'd say you are 100% correct. In both Linux and Unix (and in
> fact in any Posix compatible OS) a file "in use" is not referenced
> by any file system entry, instead a file unit id is used - which is
> nothing more nor less than a handle established when the file is
> opened by a program and which lives on until the file is closed by
> the same program. Hence the behaviour you describe.

Thanks Noons.

Just to clarify one stage on, though.

Suppose I alter a symbolic link so that it points to directory B (instead of directory A) precisely as an archive is actually being written (and hence the link is in use)?

I thought, and you seem to agree, that the entire archive will end up safe and sound in destination A. But, when that archive is finished, does the link become 'unused', and hence the change to it takes effect? Would that mean the *next* archive gets sent to destination B?

Or is the symbolic link 'grabbed' by Oracle at instance startup, so that it is always 'in use', even when no archiving is physically taking place?

If so, because the link is always 'in use', then it's pointless the OP using the symlink in his init.ora. But if the alteration can kick in without an instance bounce, albeit only for the next archive, then I can see a use for it.

Regards
HJR Received on Thu Oct 21 2004 - 15:28:09 CDT