Hi Christian,

>We have to make normal file-backups. These files are lateron copied over
>network to another server (Windows NT) and backuped via a backup-tool in a
>tape-library. Yes, I know: not very beautiful ;-)

Which backup tool, by the way?

> I need an
> automatical backup (e.g. every weekend a fullbackup). So I 
> don't want to be
> prompted for a new filename... I've setup a media-group with about 10
> backup-devices. They're parallel filled but not for 
> performance reasons.
> Only to avoid the prompt for a new filename...

OK, three solutions come to mind:

1. How about using dbmcli, file x with prepared "backup_replace
..."-commands and a "dbmcli ... <x"? The luxurious version would be a script
talking with dbmcli and thereby avoiding error messages for the
backup_replace-commands after the end of the backup.

2. You could also use a dd to read the backup data from a pipe and write it
to disk. Provided, that you have a dd, which can handle files larger than 2
GByte. Keep in mind to create to pipe before it should be used.

3. You can use the program pipe2file delivered with SAP DB. While probably
not able to break the 2Gig barrier on Linux, it can cut the backup data into
a lot of small pieces. So backup again into a pipe and start "pipe2file -d
pipe2file -f <filename> -p <pipename> -s <filesize> -nowait". After the
backup you should find a lot of files of size <filesize> named <filename>.0,
<filename>.1, ... . With "pipe2file -d file2pipe -f <filename> -p <pipename>
-s <filesize> -nowait" you would recreate the stream of backup data into the
pipe <pipename>.

HTH,
Tilo Heinrich
SAP Labs Berlin

_______________________________________________
sapdb.general mailing list
[EMAIL PROTECTED]
http://listserv.sap.com/mailman/listinfo/sapdb.general

Reply via email to