Forum OpenACS Q&A: OT: Oracle off-line export

Collapse
Posted by Andrei Popov on
Sorry for an off-topic message, but maybe someone (of Oracle gurus) can give me a hand with this one: I have a backup of an Oracle table space, as in just plain files (.dbf's) copied on a tape.  I want to retrieve data from them (as I cannot re-mount them).  I have heard of a utility that can do that.  Does anybody no anything to this extent or am I just dreaming?

Alternative suggestions are also welcome.

Thanks.

Collapse
Posted by Glenn Cadman on
Whilst I may be shot down in flames, I dont think it is possible to do as you wanted. The system tablespace contains the data dictionary, which is needed to retrieve the data logically. Also the existing datafiles and the backup datafiles are now inconsistent, (unless it was an archive tablespace).

To restore I think you have to shutdown and save the current db, restore the old db and tablespace, export your old data. The move back your good database and startup again, import the good data.

Collapse
Posted by Andrei Popov on
<blockquote> To restore I think you have to shutdown and save the
current db, restore the old db and tablespace, export
your old data. The move back your good database and
startup again, import the good data.
</blockquote>

This is a luxury I no longer have -- the rest of the db is
gone...

I remember seeing a mention of a certain utility (dbu?
dbunload?) that could do something along the lines of what I
need, but I cannot seem to find a single trace of it
anywhere...

Collapse
Posted by Sebastiano Pilla on
Sometimes I've seen mentioned an Oracle DUL (data unloader) tool, to be used for situations like yours. However, DUL seems to be an internal Oracle tool: you'll have to call Oracle support and ask to have somebody sent to your place to recover the database, and it can be very expensive.

You could also try http://sourceforge.net/projects/jdul/, although the project doesn't seem to have any release now.

All the best, take it as a lesson in backup and recovery you'll never forget.

Collapse
Posted by Glenn Cadman on
That must be one great tool when it comes out. Just thinking about it, without a central datadictionary, logically sorting out the bits of each table from each other across mutiple datafiles, chained rows etc would be a link putting together of a 20000 piece jigsaw puzzle.

However logically dumping the contents (row level) of a datafile to a flat file then letting a human sort it all out, may be possible.

Collapse
Posted by Andrei Popov on
Thanks y'all.  A pitty it is, then...  40 gigs of data... lost... :(