Re: Question on orasrv

>
>In <RWESSMAN.93Oct28145021_at_rwessman.us.oracle.com> rwessman_at_rwessman.us.oracle.com (Rick Wessman) writes:
>
>
>>The problem is that on UNIX boxes, "shutdown abort" doesn't really kill
>>the oracle shadow processes. Instead, it just removes the semaphores and
>>relies on the processes dying when they try to latch a
>>resource. However, the background processes do exit.
 

>>So, until the process attempts a latching operation it is free to do
>>operations on behalf of the user.
>
>[ stuff deleted ]
>
>I would have expected this query to require a latching operation.
>Unless the whole table is already in the cache, surely Oracle would need
>to acquire at least one latch in order to read data into a
>DB_BLOCK_BUFFER.
>
>Or are you saying that this behaviour depends on sch_ft_trans being 100%
>cached?
>
>graeme
Yes. The occurrence of this phenomenon depends on whether a latch is required. If a resource is cached and the operation is read only, it will succeed.

Please note that this behavior is a bug and should not be relied upon.

                                Rick Received on Mon Nov 08 1993 - 17:10:30 CET