Re: Oracle Security Alert for CVE-2012-1675 - 10g extended support
Date: Sun, 6 May 2012 07:34:25 +0200
Message-ID: <CALH8A91mhqM5G1U47rX-e_k6QstPsMRDqXn1o3qm_=yzHPwd9Q_at_mail.gmail.com>
In a RAC environment that should not be a big issue (still there is a little 'gap').
Within 3 seconds (that is the 'gap') PMON tells all listeners, if the status of any other listener (it cares about) changed. So the SCAN listeners mark it as 'state:blocked' in 'lsnrctl services' and does not redirect any connections to it anymore. (again one more advantage of SCAN setup)
I have done a similar test in more details here: http://berxblog.blogspot.com/2012/05/who-cares-if-listener-is-dying.html
hth,
Martin
On Sat, May 5, 2012 at 10:29 AM, Martijn Bos <maboc_at_maboc.nl> wrote:
> Hi,
>
>>
>> In your (and everyone else's experience), how long does a listener
>> start/stop take? If it's in a shell script it might be very quick, and
>> done over a quiet period shouldn't hurt too much.
>>
>
> In my experience stopping/starting a listener doesn't take more then a few seconds.
> However, it can take a while (have no figures available) before the "locallistener databases" will register them selfs with the listener.
>
> The other day it took about 30 seconds for the last local listener to register, in out test environment.
>
> The existing connections are not impacted, only new connections can't be made. So I gues that applications which user some kind of connection-pooling mechanism are not hurt, since they do not (generally) need to create new connections all the time. Applications whitout connection pools may suffer from "database loss" or "no database" for a while. I guess it's up to "the bussiness" to see whether that is acceptable.
>
> Best Regards,
> Martijn
>
>
>
>
> --
> http://www.freelists.org/webpage/oracle-l
>
>
-- Martin Berger martin.a.berger_at_gmail.com Lederergasse 27/2/14 +43 660 660 83306 1080 Wien http://berx.at/ -- http://www.freelists.org/webpage/oracle-lReceived on Sun May 06 2012 - 00:34:25 CDT