Re: Data corruption error

In <31E43FD2.9FB_at_execpc.com> JTseng <tseng_at_execpc.com> writes:
>
>Has anyone encountered a data corruption error when running
>Oracle 7.1.6 on HP SUN UNIX?
>
>I'll get the error code but it basically comes down to
>a corruption that occurs sporadically in a rather large data
>warehouse. It has been found that the corruption is narrowed to
>a specific row in the table. Selecting up to and after the row
>(using rowid > and rowid < ) does not cause any problem, but
>any selects which involve table scans brings up the error
>apparently because Oracle passes over the corrupted part of
>the table.
>
>Oracle's answer to this is to select out of the table, drop it,
>recreate it and reload it. This would be fine for a small table,
>but for one with 5 million rows, this isn't acceptable, since the
>whole workaround takes umpteen hours. And the corrupted data
>is lost.
>
>I'm sorry I don't have the error number on hand. I'll try and repost
>with it.
>
>Thanks in advance
>
>J Tseng
>tseng_at_execpc.com

How long would it take to create and reload your 6 million table ? I use the sql*loader direct path option, Drop all constraints and indexes, and it will take about 1 hour to load 6 million rows and 15 minutes to generate 1 index using the parallel index option.

The second option is to create a table to hold 6 million rows minus the corrupted block. You will have to recreate the rows that are corrupted.

Glen Case Received on Thu Jul 11 1996 - 00:00:00 CEST