Re: v$session question

  • From: Ahbaid Gaffoor <ahbaid@xxxxxxx>
  • To: ORACLE-L <oracle-l@xxxxxxxxxxxxx>
  • Date: Wed, 04 Feb 2009 09:01:08 -0800

At one shop we got bitten using "kill -9" while an RMAN backup was in session, granted this was on 9.2.0.6 on AIX (can't recall the version), we did not kill the RMAN session but killing one of the user processes at the OS level after the backup had started crashed the db.


This turned out to be a bug, which Oracle patched.

For each of us our configs are very different, best approach I've found is to re-create the situation somewhere else and test it.

Ahbaid

Jared Still wrote:



On Fri, Jan 30, 2009 at 8:45 AM, Ram Raman <veeeraman@xxxxxxxxx <mailto:veeeraman@xxxxxxxxx>> wrote:

    Thank you Tom.
I would think that killing via oracle first is safer because
    oracle would roll back any work done. If done via OS first, will
    Oracle roll the transaction back and leave the database in a
    consistent state?  If there is no entry in the V$transaction, then
    I think killing the process from the OS should be ok. Any
    corrections to this theory?


This question, or something like it, comes up every time someone mentions
using kill -9 before using 'alter system kill'.

Using kill -9 to first kill the process has a long history of success.

In some cases, failure to use kill -9 first can result in a session that
never goes away, requiring a bounce of the database.

A google search should find several posts on this.

Jared Still
Certifiable Oracle DBA and Part Time Perl Evangelist

--
//www.freelists.org/webpage/oracle-l


Other related posts: