Re: ORA-00471: DBWR process terminated with error
Date: Thu, 22 Dec 1994 14:28:24 +0000
Message-ID: <788106504snz_at_jlcomp.demon.co.uk>
In article <3ct6eq$5dh_at_toon.ctp.com> pmuzu_at_ctp.com "Pavan Muzumdar" writes:
> In article <3ckf49$fo8_at_bmerha64.bnr.ca>, Jason Lisenchuk <hris_at_nt.com> writes:
> |> Any insight into what might be causing this? This obviously affects the
> |> integrity of REDO logs, and not surprisingly, results in a corrupted
> |> database. This error occurred on a high end HP 9000 running Oracle
> |> 7.0.16, 512 MB RAM, 12 GB disk array (half of which is free). I am
> |> assuming that either a hardware fault occurred or something went awry in
> |> the system software.
> |>
I missed the start of this thread, but the bits I have seen match a recent problem I was called to.
Hardware: HP9000, Oracle: 7.0.13, O/S HP-UX 9.04
Symptoms:
Data base crashed out with 00471 error from DBWR, and similar
numbers 473, 474 as the rest of the backgrounds came down too.
Several redo logs got smashed as this happened.
I found that the problem seemed to be a corrupted index. When I dropped and rebuilt the index, the crashes stopped happening.
The task crashing the system was a simple DELETE statement that had to delete about 10% of a large table, selecting the rows through the offending index.
I was not able to investigate in detail as there was a deadline fast approaching, so I did not have time to try any event setting or block dumping etc., but it is of course possible that the index had got corrupted as a result of the first crash, and the corruption coincidentally caused the same type of crash thereafter.
-- Jonathan LewisReceived on Thu Dec 22 1994 - 15:28:24 CET