Re: Pluggable database in 12C
Date: Wed, 24 Oct 2012 21:22:05 +1100
Message-ID: <k68fgd$vj8$1_at_dont-email.me>
Mladen Gogala wrote,on my timestamp of 24/10/2012 10:56 AM:
> On Tue, 23 Oct 2012 16:12:12 -0700, joel garry wrote:
>
>> According to Alert Log Messages: Private Strand Flush Not Complete [ID
>> 372557.1] it's just telling you the consolidation of the private redo is
>> taking some time. I would guess the suggestion of adding db writers to
>> keep the cache clean is a bit of compulsiveness. Perhaps Jonathan knows
>> better.
>
> Joel, I'm no Jonathan by any stretch of imagination, but it doesn't look
> to me that adding DBWR would particularly helpful with REDO? How would
> database block writer help LGWR consolidate the redo entries from the
> private redo strands?
> BTW, I met Jonathan in person and we don't look even remotely alike.
Not sure about the likeness. :)
But I do know that if I jack up the priority of the lgwr process the incidence
of this message is greatly reduced although not completely eliminated.
The issue I have with the "consolidation of the private redo is taking some
time" thing is simply this: which stat or wait tells me it is happening, and by
how much?
I'm fundamentally alergic to tuning by "some": either there is a numeric,
tallied base for an observation, or it's not worth the surface it's written on!
And of course: I'm not directing this at Joel! I've seen the same text in some
of the mentioned MOS docs and they indicate it means, essentially, nothing.
My gut feel is that someone forgot to remove a debug message from somewhere in the 11gr2 code and now we have this thing showing up all over the place and meaning essentially nothing!
And there are more. In fact, the 11gr2 alert file is NOTORIOUS for quite a few comments with no WARNING or ERROR tag before them, or any ORA number. Which makes it impossible to accurately make any sense of them, other than by coincidence! Received on Wed Oct 24 2012 - 12:22:05 CEST