Re: 10g datapump

From: Chuck <chuckh1958_nospam_at_gmail.com>
Date: Mon, 20 Oct 2008 21:24:28 GMT
Message-ID: <gG6Lk.2482$Rx2.1551@nwrddc01.gnilink.net>


gazzag wrote:
> On 14 Oct, 16:29, Chuck <chuckh1958_nos..._at_gmail.com> wrote:

>> gazzag wrote:
>>> On 13 Oct, 21:40, Chuck <chuckh1958_nos..._at_gmail.com> wrote:
>>>> When exporting to multiple files using options like filesize=100m and
>>>> dumpfile=filename%U, is there a way to definitively know when oracle is
>>>>  finished writing to a specific dump file?
>>>> Here's why I ask. I need to compress the dump files, but don't want to
>>>> wait until the entire job is finished before starting. I want to begin
>>>> compressing files once I know oracle is finished with them, while it may
>>>> still be writing to other files.
>>> Any particular operating system and version?  Actually, a proper
>>> Oracle version might help too.
>> Oracle 10.2.0.3 on Solaris 10

>
> As I understand it 10gR2 supports a COMPRESSION parameter with
> DataPump as well as a PARALLEL one. The functionality that you're
> trying to create might actually exist. Does this document help?
>
> http://www.oracle.com/technology/products/database/utilities/pdf/datapump11g2007_quickstart.pdf
>
> Failing that, a home-grown solution might unvolve the Unix "fuser"
> command.
>
> HTH
>
> -g

That document mentions the compression option only on the 11g datapump utility, when connected to an 11g database.

I tried both the fuser and lsof commands. It seems like the 10g expdp closes files that it's not completely done with and reopens them later. I've still yet to find a way to determine if expdp is really done with a file and is not going to go back to it later. I suspect there might be a query you could run against the master table for the export, but there's no documentation on how that file is used. Received on Mon Oct 20 2008 - 16:24:28 CDT

Original text of this message