Re: Corrupt Primary Key index on table causing ORA-600 internal error

From: Mark D Powell <Mark.Powell_at_eds.com>
Date: 1 Aug 2004 18:27:08 -0700
Message-ID: <2687bb95.0408011727.27fe2580_at_posting.google.com>


biwombi_at_hotmail.com (Thiko) wrote in message news:<b8d3b40c.0408011124.6283acac_at_posting.google.com>...
> Hi
>
> I have a corrupt PK index on a table column. It is a unique PK. It
> needs to be dropped and recreated to cure the corruption.
>
> The table is on a backup database which is in replication having
> records added to it to keep the two databases in sync.
>
> Am I correct in think that if I:
>
> 1. Suspend the replication so no data inconsistency happens on the
> table while the PK is disabled.
> 2. Disable the corrupt PK on the table - then the index will get
> dropped?
> 3. Enable the PK on the table - and the index will get rebuilt with
> the corruption gone?
> 4. Re-enable the replication.
>
> All will be fine? The table has over 9million rows so I realise it
> will take a while to re-enable the primary key but as is is a backup
> database in replication performance to users won't be affected.
>
> Many thanks!

Thiko, dropping and re-creating the PK will probably correct the situation since from your information it appears to be just a corrupted index. Depending on the type of replication in use and hardware capacity re-creating the PK could be a 30 second to 10 minute operation so turning of the replication may not even be necessary if you are on the low end of the scale.

I would think a complete refresh might be an option.

By the way when a newsgroup has subgroups you should generally post in the appropriate subgroup and not in the base group.

HTH -- Mark D Powell -- Received on Mon Aug 02 2004 - 03:27:08 CEST

Original text of this message