Re: kill session questions

In article <QKT06.1471$F_6.409228_at_news.uswest.net>,   "Verna Legaspi" <vlegaspi_at_uswest.net> wrote:
> Hello -
>
> There seemed to be a procedure gone awry and I had to do a
> ALTER SYSTEM KILL SESSON 'sid, serial#'
>
> on 2 sessions. They were both ACTIVE when I tried to kill them.
 However,
> it's been at least 3 hours, now - and they're both still simply
 flagged as
> "to be killed."
>
> Is this not uncommon? Is there any other way for me to force it to be
> killed completely? Also - am I correct in assuming none of their
 resources
> or locks/blocks have been released, yet? ... at least not until they
 finally
> fall of the v$session view?
>
> TIA,
> Verna
>
> Hi Verna,

In my opinion you're correct. The KILL SESSION command will cause PMON to rollback the user transaction,release locks and free resources upon execution. As to active sessions, I had a quick look at a book here but it's an ambiguous answer..it says that if the user session is making an SQL call to the server (i.e. it's active), the transaction is rolled back and the user immediately receives this message:

ORA-00028: your session has been killed. (Oracle's usual rosy type of message)..
Here's the ambiguous part..."If the user session is performing some activity that must be completed and cannot be interrupted, the Oracle server waits for this activity to complete."

I hope this is of some help Verna.

Steve

Sent via Deja.com
http://www.deja.com/ Received on Sat Dec 23 2000 - 06:18:42 CST