Re: Disappearing Sessions - Part 2

From: Anurag Varma <avoracle_at_gmail.com>
Date: Fri, 8 Feb 2008 15:00:29 -0800 (PST)
Message-ID: <bb46a7ae-ae01-4948-a938-1ca5d866ce81@i12g2000prf.googlegroups.com>


On Feb 8, 5:07 pm, "Dereck L. Dietz" <diet..._at_ameritech.net> wrote:
> Oracle 10.2.0.3.0
> Windows Server 2003
> 16 GB of memory
>
> Our off-site DBA had this to say about the disappearing sessions and
> ORA-07445 errors:
>
> "I think the problem with disappearing sessions is the dispatcher process.
>
> when we increased the session limit, I did not take into account that there
> are web connections coming in also.
>
> ORA-07445 is defenitely points back to Hardware/OS. Since the memory
> upgrade (16 which will need to go to 32), we have not seen any errors."
>
> Does this make sense when:
>
> 1. I'm running my process in Dedicated Server mode, not Shared Server.
>
> 2. If I add just one more nested table that gets loaded in my code my
> process will disappear however if
> I take it out my process will run to completion.
>
> 3. If I force a full table scan my process will complete. If I try using a
> FIRST_ROWS hint or change
> my query to use an extra column join (which seems to cause a hash join) my
> process disappears.
>
> I should also state that just today, after receiving this information from
> our off-site DBA that the database went down (around the time it went down
> there looked like there was massive contention memory-wise) and we have a
> new ORA-07445 error.

ORA-7445 refers to an Oracle internal error. The arguments provided in that error
might point to an existing bug. The DBA should open a SR with oracle (similar to when a
ORA-600 error occurs).
The root cause can be anything ..
What you describe sounds like a bug.

Anurag Received on Fri Feb 08 2008 - 17:00:29 CST

Original text of this message