Leyne, Sean wrote:

> Mark,
>
>>>> all provided here and there 10 days ago. And I confirm the simple test
>>>> proves it. It's a show stopper when accessing the MON tables in
>>>> production.
>>> While I agree that the test does highlight the problem, I think the test
>> case is a little extreme -- continuously querying the MON$ tables over and
>> over...
>>> This suggests an abuse of what the MON$ tables were designed for.
>> I think you are confusing a reproduction recipe or testcase that clearly
>> illustrate the problem with actual usage where the problem also exists,
>> but is less visible in the short term.
>
> I realize, now, that my message could be read to apply to the test case.
>
> However, I was referring to the original report which found the condition as a result of continuously querying the MON$ to monitor the server state/performance.

On the other hand, repeatedly querying statistics views/tables is
exactly the same usage as for instance most tools in Oracle use to show
server statistics or query specific statistics (ie certain server or
query overviews in TOAD refresh every 120 seconds if opened).

I have also ran various loadtests on large systems where every 5 or 10
seconds the statistics of the Oracle servers where queried to be able to
correlate the load applied on the application with statistics on the
various servers involved.

I don't think this specific case is that out there for production usage
(or in my example: test environment usage). Especially not if applied
from Nagios or other monitoring tools.

Mark
--
Mark Rotteveel