Re: dynamic service registration

"srivenu" <srivenu_at_hotmail.com> wrote in message news:1a68177.0403212323.1ec39d24_at_posting.google.com...
> The status UNKNOWN will show for any service that was created based on
> SID_LIST in the listener.ora.
> These such services do not have the functionality to report status to
> the listener. The newer method of service registration with the use of
> PMON will in fact create service handlers for the database that do in
> fact report thier status on a regular basis to the listener process.
> Since the older SID_LIST handlers were not coded with that
> functionality they simply report as UNKNOWN in the listener display.
> So its nothing to be alarmed by its simply the older code cannot do
> what the newer code can.
> The SID_LIST section only needs to be used for instances such as
> PLSExtProc.
> It is not normally a problem having the instance duplicated in the
> SID_LIST, but it is preferrable not to have it.

That, and the sentence before it, is not quite true. The Intelligent Agent isn't. Intelligent, that is. It only "discovers" those services which are statically declared in the SID_LIST section of the listener.ora, and will not discover services which are using dynamic instance registration. Therefore, if you intend to manage your 8i and 9i databases with Enterprise Manager, it is practically compulsory to have both DIR and static SID_LIST declarations for a single instance.

> One of the reasons why Oracle recommends not using one, is because it
> may cause un-necessary overheads on the listener process.
>
>
> The PMON registration cycle is every 1 minute.

I'd be interested to see your source for that statement. It's very difficult to pin down exactly what PMON's registration schedule is, because PMON has other things to do. But looking in O_H/network/log/listener.log, I can see registration intervals of 1 minute, 5 minutes, 10 minutes and 20 minutes -and indeed of most other time intervals besides. For example:

22-MAR-2004 21:08:30 * service_update * lx92 * 0
22-MAR-2004 21:11:51 * service_update * lx92 * 0
22-MAR-2004 21:12:57 * service_update * lx92 * 0
22-MAR-2004 21:13:30 * service_update * lx92 * 0
22-MAR-2004 21:14:36 * service_update * lx92 * 0
22-MAR-2004 21:15:42 * service_update * lx92 * 0
22-MAR-2004 21:16:48 * service_update * lx92 * 0
22-MAR-2004 21:19:00 * service_update * lx92 * 0
22-MAR-2004 21:19:33 * service_update * lx92 * 0
22-MAR-2004 21:29:36 * service_update * lx92 * 0
22-MAR-2004 21:49:42 * service_update * lx92 * 0
22-MAR-2004 22:09:48 * service_update * lx92 * 0
22-MAR-2004 22:50:00 * service_update * lx92 * 0
22-MAR-2004 23:00:03 * service_update * lx92 * 0
22-MAR-2004 23:30:12 * service_update * lx92 * 0
22-MAR-2004 23:40:15 * service_update * lx92 * 0
23-MAR-2004 00:50:36 * service_update * lx92 * 0
23-MAR-2004 01:00:39 * service_update * lx92 * 0
23-MAR-2004 02:41:09 * service_update * lx92 * 0
23-MAR-2004 03:01:15 * service_update * lx92 * 0
23-MAR-2004 03:11:18 * service_update * lx92 * 0

3.5 minutes between the first two lines. 1 minute intervals for the next 5 lines. 2.5 minutes for the next one. 33 seconds for the next one. Then 10 minutes, 20 minuts, 10 minutes, 40 minutes, 10 minutes, 30 minutes, 10 minutes, 10 minutes, 10 minutes, 41 minutes, 20 minutes, 10 minutes.

If I had to make a wild stab in the dark, I'd say the schedule was more generally 10 minutes rather than anything else. Though the average for this 6 hour period was 18 (call it 20) minutes.

I won't bore you with the output unless someone demands to see proof, but checking the 12 hours from 9am yesterday to 9pm yesterday, my PMON registered 139 times. Which, give or take my abysmal maths, is an average registration interval of ... 5 and a half minutes.

Now, that's Windows 9i Release 2, and it might be different on different platforms and versions. But 1 minute intervals I am pretty certain it is definitely not.

Regards
HJR
>Along with the instance
> information, it also updates the listener with the Dispacthers load.
Received on Mon Mar 22 2004 - 13:02:45 CST