RE: Setting db_32k_cache_size breaks

From: Mark W. Farnham <>
Date: Sun, 23 Nov 2008 02:54:48 -0500
Message-ID: <>

While it is quite correct for the Oracle program to complain about a parameter setting out of range, the rest of the behavior reported is certainly a serious bug.

In addition to reporting the error, Oracle should not obey the errant command, leading to the subsequent errors in attempting to re-connect.

It looks as if someone intended to print out the valid range arguments in the error template and botched that as well. If the [..] were indeed replaced with "for <OS name> is <min_block_size> to <max_block_size>" with the <> values filled in, you wouldn't have had to wait for Greg to look it up and reply.

So asking that Oracle support 32k block sizes for Linux would be an enhancement request, while the bug to report is the lack of resilience and incomplete error message.



-----Original Message-----
From: [] On Behalf Of Greg Rahn
Sent: Sunday, November 23, 2008 12:30 AM To:
Cc: Andre van Winssen; oracle-l
Subject: Re: Setting db_32k_cache_size breaks

Linux does not support a 32k block size. 16k is the max. This is why you get:

ERROR at line 1:
ORA-02097: parameter cannot be modified because specified value is invalid ORA-00382: 32768 not a valid block size, valid range [..]

On Sat, Nov 22, 2008 at 1:59 PM, Stefan Knecht <> wrote:
> I filed an SR this afternoon.
> Are you on 32-bit linux as well ?

Greg Rahn

Received on Sun Nov 23 2008 - 01:54:48 CST

Original text of this message