RE: RMAN restore into nearly full file system
- From: <Joel.Patterson@xxxxxxxxxxx>
- To: <big.dave.roberts@xxxxxxxxxxxxxx>, <charlottejanehammond@xxxxxxxxx>
- Date: Tue, 24 Aug 2010 10:54:42 -0400
You can check Metalink Note 563427.1 about RMAN 'compression' as this
note relates to sparse files concepts and the versions of oracle.
Joel Patterson
Database Administrator
904 727-2546
________________________________
From: oracle-l-bounce@xxxxxxxxxxxxx
[mailto:oracle-l-bounce@xxxxxxxxxxxxx] On Behalf Of David Roberts
Sent: Monday, August 23, 2010 6:11 PM
To: charlottejanehammond@xxxxxxxxx
Cc: oracle-l@xxxxxxxxxxxxx
Subject: Re: RMAN restore into nearly full file system
Were any of the data files on the file system 'sparse'?
Sparse files occur when disk blocks are empty, and rather than
allocating a block on disk, the inode makes a note that the block is
empty.
In the past no Oracle database files should be sparse, although locally
managed temporary tablespaces are now created as sparse files I believe.
However these will not normally be backed up by RMAN and shouldn't be a
problem.
However I believe that the AIX backup and restore utilities will create
sparse files if they have ever been used to restore or move the database
files.
Dave
On Mon, Aug 23, 2010 at 9:49 PM, Charlotte Hammond
<charlottejanehammond@xxxxxxxxx> wrote:
Hi All,
Recently, when I ran an RMAN restore I got a "no space left on device"
error and
RMAN aborted. Why did the restored data files no longer fit onto the
disk from
which they were originally backed up?
(No files had been added to the file system since the backup was taken).
ls showed the same number of bytes before and after the restore. I
presume the
problem was with "hidden" space usage such as inode blocks or file
indirect
blocks. There was only about 20Mb free on the 80Gb file system, yet 20Mb
still
seems quite a lot of degredation in the efficiency of file system
metadata
(there were only 5 data files to track). How can I explain the loss of
20Mb
between backing up and restoring the files? (The file system is JFS2,
block size
was 4k, OS is AIX, Oracle 9i (don't ask!))
I want to use this as further evidence to try to deter a manager from
routinely
overfilling file systems but I'd really like to be able to explain it
better
before doing so (he'll be asking questions!).
Many thanks!
Charlotte
--
//www.freelists.org/webpage/oracle-l
Other related posts:
