[Novalug] OT: newbie oracle question

[Kevin Kitts]

I've been working with Oracle for many years - and it is pretty unusual for
a table to become corrupted. On the very few times I've seen something like
this there was a hardware fault of some kind going on. Once, some data
placed on tape was corrupted when we got it off the tape and used it to
build a database. Incidentally, in the case above, only the index was
corrupted - not the actual data - so I just dropped the index and rebuilt
it.

As others have mentioned the Oracle alert log is the first place to look for
errors. Also, you can enter the following on a command line:

oerr ORA xxxxx

where xxxxx is the error number - and the oerr program/facility will give
you both the cause of the error and a recommended solution. It is often -
but not always - helpful. I hate it when it says "Contact your Oracle
DBA"... ;-) Doh....

Good Luck,

Kevin

On 10/22/07, Peter Larsen <plarsen at famlarsen.homelinux.com> wrote:
>
> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
>
> Zakai Kinan wrote:
> > I have oracle 9iR2 installed in a lab and a table got
> > corrupted on Friday.  I went ahead and restore the
> > whole database.  Now I am realizing that I can't find
> > a way to diagnose the problem.  Is this a fact?  My
> > research has not turned up anything yet.  I am a noob
> > in oracle.  I have 9iR2 installed for a specific
> > project.  Any insight would be helpful
> Very little information indeed.
> How do you know the "table" is corrupted? What error did you receive,
> and how did you go about to restore the table?
>
> As cliff points out, the alert log is a good place to start for system
> issues, but you'll ALWAYS get an error on the console. If you have
> serious system issues, you most likely get one of the generic codes like
> ORA-600; and more information can then be found in the alert and trace
> logs.
>
> Btw. if it's a simple/small databsae try the 10XE version. While not a
> complete Oracle database, it's great for smaller systems; and it's free :)
>
> Once you know the error, that'll hint at the cause. If there is media
> corruption, you'll deal with the media issue; if you have an application
> bug you'll deal with that; or if your datamodel is bad and allows
> inconsistency you'll have to deal with that. All in all, these and a ton
> of other reasons for data being lost. So a little more information is
> needed. Start with the error you got and the operation you were doing;
> then proceed to the primary logfiles (alert log) and from there to the
> trace files. However alert/trace are only used in serious cases. When it
> comes to application issues, or datamodel issues - there won't be
> anything in the alert logs.
>
> Good luck :)
>
> --
>    Peter Larsen
> -----BEGIN PGP SIGNATURE-----
> Version: GnuPG v1.4.7 (GNU/Linux)
> Comment: Using GnuPG with Fedora - http://enigmail.mozdev.org
>
> iD8DBQFHHUNwHbLDna3OFD0RAnpSAKCOM2pKjpgYWwQVvjYrdYaKHwQRxACgjXML
> mD+VGKdudRR06R5JoJfcjnM=
> =878p
> -----END PGP SIGNATURE-----
> _______________________________________________
> Novalug mailing list
> Novalug at calypso.tux.org
> http://calypso.tux.org/cgi-bin/mailman/listinfo/novalug
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://calypso.tux.org/pipermail/novalug/attachments/20071023/7bceb67c/attachment.html