Re: Import error between versions

> My production system is moving from 8.1.5 on Solaris7 to 9.2.0.4 on Linux. I
> exported my users into dump files of 1GIG each. I can import these fine
> either into new users on the same system or into a new system running the
> same OS and DB version so i know the dumps are not corrupt.
>
> The issue arises when i try and import these into a new system running a
> different OS and DB version. The import goes OK until the point it would
> appear to move onto the second file in the dump sequence. The dump fails and
> the error reported just says segmention fault. There is nothing logged into
> the alert log or any other log. It behaves like this if i import into 8.1.7
> on Solaris or Windows or also 9.2.0.4 on Linux. The even 'odder' thing is
> that if i allow the export to run into a single file, so avoid the FILESIZE
> parameter, it imports fine into all DB versions...
>
> Has anyone seen anything like this before or have any idea what might cause
> it?
>
> My exports are scripted each night and have the following syntax...
>
> exp xxxx/password_at_INST FILESIZE=1G
> FILE=$BACKUPDEST/xxxx0.dmp,$BACKUPDEST/xxxx1.dmp,$BACKUPDES
> T/xxxx2.dmp,$BACKUPDEST/xxxx3.dmp,$BACKUPDEST/xxxx4.dmp,$BACKUPDEST/xxxx5.dm
> p,$BACKUPDEST/xxxx6.dmp,$BACKUPDEST/xxxx7.dmp,$BACKUPDES
> T/xxxx8.dmp,$BACKUPDEST/xxxx9.dmp log=$BACKUPDEST/xxxx.log

I'm starting to think this is an OS thing - through a few responses it seems that when you move OS's obviously the checksum of the fileschanges. This means that oftem the OS can't find the EOF marker and so quits. This would explain why if i dump everything to a single file i can get it to work.

Can i ask the group how they manage to export datasets that are going to be bigger than the filesize limit of the filesystem between OS's?

thanks

Tom Received on Tue Nov 18 2003 - 05:26:11 CST