Prev: Export datapump has stopped woring
Next: ORA-30036 during ALTER TABLE SHRINK
From: John Hurley on 19 May 2010 08:52
zigzag:

# This Oracle is really buggy. I did not notice while I was testing
for last 3 months, but as soon as I went to production, I have
encountered this problem. Now my scripts for shuttiong down database
etc will not work because they will hung.

You did not notice a problem starting up and shutting down databases
during 3 months of testing?

We don't run anything on hpux anymore ... all linux except for one
solaris box. In the old days Oracle maintenance on hpux used to be
solid.

There are a number of platform specific issues I think with some of
the platforms and patchset updates that you need to look at and review
in the available Oracle doc for each of the patchset updates.

That being said ... I would not be surprised if it was something else
besides Oracle maintenance that is messed up in your environment.

From: Mark D Powell on 19 May 2010 09:47
On May 19, 8:52 am, John Hurley <hurleyjo...(a)yahoo.com> wrote:
> zigzag:
>
> # This Oracle is really buggy. I did not notice while I was testing
> for last 3 months, but as soon as I went to production, I have
> encountered this problem.  Now my scripts for shuttiong down database
> etc will not work because they will hung.
>
> You did not notice a problem starting up and shutting down databases
> during 3 months of testing?
>
> We don't run anything on hpux anymore ... all linux except for one
> solaris box.  In the old days Oracle maintenance on hpux used to be
> solid.
>
> There are a number of platform specific issues I think with some of
> the platforms and patchset updates that you need to look at and review
> in the available Oracle doc for each of the patchset updates.
>
> That being said ... I would not be surprised if it was something else
> besides Oracle maintenance that is messed up in your environment.

Is your production system on the same server as your testing? If not,
I would go back and compare the OS settings to those recommended by
Oracle in the platform specific installation guide. The 11g system I
had on Linux started every bit as fast as my 9i an 10g systems on
AIX. My employer has Oracle 11g on HP-UX and I have not heard any
complaints from the DBA's working with those systems.

After double checking the OS settings as John and I recommend you
should also double check your spfile database parameters settings and
compare those to what you used in test. It could be what you are
experiencing may result from a specific combination of parameters that
most of use have not used in the fashion your environment is
configured for.

HTH -- Mark D Powell --
From: joel garry on 19 May 2010 14:08
On May 18, 2:15 pm, zigzagdna <zigzag...(a)yahoo.com> wrote:
> On May 18, 9:50 am, ddf <orat...(a)msn.com> wrote:
>
> > On May 18, 8:15 am, vsevolod afanassiev
>
> > <vsevolod.afanass...(a)gmail.com> wrote:
> > > Could be related to memory management: Oracle 9 and 10 create only one
> > > shared meemory segment on startup, Oracle 11 may create many segmeents
> > > in order to allow dynamic SGA resizing.
>
> > Not necessarily.  9i did not use NUMA for shared segments, although it
> > did use NUMA for private memory areas. 10gR2 changed that and
> > configured both shared and private memory for NUMA which results it
> > multiple shared memory segments created at startup.  This depends upon
> > your hardware configuration and  O/S (both must be NUMA-enabled) but
> > without setting additional parameters or applying a patch 10gR2 can
> > start multiple shared memory segments and you may not know it's being
> > done.
>
> I found a bug in Oracle 11g today related to startup,
> I did startup
>
> It said Oracle instance started
> Shwhed me sga information
> SHowd Database Mounted
>
> After mounted it got hung, in the past after 10 minute sor so, it will
> show I thing database open, but today it did for hours.
>
> I thought my database was not opened. then my users told that they are
> accessing database without problem,
>
> So, I opened another problem and  wow I could connect so database was
> open eventhough it wa shung in startup.
>
> HAs anyone notice dthis with Oracle 11.1.0.7.1
>
> This Oracle is really buggy. I did not notice while I was testing for
> last 3 months, but as soon as I went to production, I have encountered
> this problem.
>
> Now my scripts for shuttiong down database etc will not work because
> they will hung.

I'm wondering whether you are running into something dbcontrol
related. Are you seeing lots of cpu usage (see MOS 738115.1)? Are
you using the command line to run things, or doing them through
dbconsole? Who is using the cpu?

If what you are seeing doesn't match what the users are seeing, maybe
you are being misled.

jg
--
@home.com is bogus.
http://gizmodo.com/5542252/men-floating-in-bouncy-castle-interrupt-international-sailing-regatta
From: zigzagdna on 19 May 2010 23:06
On May 19, 2:08 pm, joel garry <joel-ga...(a)home.com> wrote:
> On May 18, 2:15 pm, zigzagdna <zigzag...(a)yahoo.com> wrote:
>
>
>
>
>
> > On May 18, 9:50 am, ddf <orat...(a)msn.com> wrote:
>
> > > On May 18, 8:15 am, vsevolod afanassiev
>
> > > <vsevolod.afanass...(a)gmail.com> wrote:
> > > > Could be related to memory management: Oracle 9 and 10 create only one
> > > > shared meemory segment on startup, Oracle 11 may create many segmeents
> > > > in order to allow dynamic SGA resizing.
>
> > > Not necessarily.  9i did not use NUMA for shared segments, although it
> > > did use NUMA for private memory areas. 10gR2 changed that and
> > > configured both shared and private memory for NUMA which results it
> > > multiple shared memory segments created at startup.  This depends upon
> > > your hardware configuration and  O/S (both must be NUMA-enabled) but
> > > without setting additional parameters or applying a patch 10gR2 can
> > > start multiple shared memory segments and you may not know it's being
> > > done.
>
> > I found a bug in Oracle 11g today related to startup,
> > I did startup
>
> > It said Oracle instance started
> > Shwhed me sga information
> > SHowd Database Mounted
>
> > After mounted it got hung, in the past after 10 minute sor so, it will
> > show I thing database open, but today it did for hours.
>
> > I thought my database was not opened. then my users told that they are
> > accessing database without problem,
>
> > So, I opened another problem and  wow I could connect so database was
> > open eventhough it wa shung in startup.
>
> > HAs anyone notice dthis with Oracle 11.1.0.7.1
>
> > This Oracle is really buggy. I did not notice while I was testing for
> > last 3 months, but as soon as I went to production, I have encountered
> > this problem.
>
> > Now my scripts for shuttiong down database etc will not work because
> > they will hung.
>
> I'm wondering whether you are running into something dbcontrol
> related.  Are you seeing lots of cpu usage (see MOS 738115.1)?  Are
> you using the command line to run things, or doing them through
> dbconsole?  Who is using the cpu?
>
> If what you are seeing doesn't match what the users are seeing, maybe
> you are being misled.
>
> jg
> --
> @home.com is bogus.http://gizmodo.com/5542252/men-floating-in-bouncy-castle-interrupt-in...- Hide quoted text -
>
> - Show quoted text -

I was able to reproduce same problem in bboth test and production
now, but in last 3 months when I was tetsing problem never occured,
I created a ticket with Oracle, have not yet ggotten any response.

I am not being mislead. After mounting startup was hung (I not showing
any output), I was running it in sqlplus. From other sqlplus I could
connect to database without problem, so my users who are running third
party application were running fine while Oracle startup was keeping
me in suspense.
From: Mladen Gogala on 20 May 2010 00:08
On Wed, 19 May 2010 20:06:02 -0700, zigzagdna wrote:


> I was able to reproduce same problem in bboth test and production now,
> but in last 3 months when I was tetsing problem never occured, I created
> a ticket with Oracle, have not yet ggotten any response.
>
> I am not being mislead. After mounting startup was hung (I not showing
> any output), I was running it in sqlplus. From other sqlplus I could
> connect to database without problem, so my users who are running third
> party application were running fine while Oracle startup was keeping me
> in suspense.

The only thing that you can do is to trace the startup with strace or
truss or whatever there is on HP-UX. That should tell you where the time
is spent. Opening a SR is a good move, too. Make sure that there is a
decent distance between your system and the LHC, because there may be a
miniature black hole devouring the output from your sqlplus session.
As long as your system starts before 12/21/2012, you should be fine.



--
http://mgogala.byethost5.com
First  |  Prev  | 
Pages: 1 2 3
Prev: Export datapump has stopped woring
Next: ORA-30036 during ALTER TABLE SHRINK