I have 10 oracle instances runnig on IBM AIX 5.1 P6000 server that has
8GB of RAM and 4GB of Swap. Last weekend the server became
inaccessible all the services were halted and we had to restart it.
We suspct that it was excesive paging that caused the problem.
How can we prevent it from happening again.

Thank you

"DBA" <marcelmix@yahoo.com> a écrit dans le message de
news:3607af43.0404061304.4c4e749f@posting.google.c om...
> I have 10 oracle instances runnig on IBM AIX 5.1 P6000 server that has
> 8GB of RAM and 4GB of Swap. Last weekend the server became
> inaccessible all the services were halted and we had to restart it.
> We suspct that it was excesive paging that caused the problem.
> How can we prevent it from happening again.
>
> Thank you

Use one instance instead of 10.

Regards
Michel Cadot

Michel Cadot wrote:
> "DBA" <marcelmix@yahoo.com> a écrit dans le message de
> news:3607af43.0404061304.4c4e749f@posting.google.c om...
>
>>I have 10 oracle instances runnig on IBM AIX 5.1 P6000 server that has
>>8GB of RAM and 4GB of Swap. Last weekend the server became
>>inaccessible all the services were halted and we had to restart it.
>>We suspct that it was excesive paging that caused the problem.
>>How can we prevent it from happening again.

1) init 1
2) Add more memory
3) lsnrctl stop

"Michel Cadot" <micadot{at}altern{dot}org> wrote in message news:<40732603$0$26450$626a14ce@news.free.fr>...
> "DBA" <marcelmix@yahoo.com> a écrit dans le message de
> news:3607af43.0404061304.4c4e749f@posting.google.c om...
> > I have 10 oracle instances runnig on IBM AIX 5.1 P6000 server that has
> > 8GB of RAM and 4GB of Swap. Last weekend the server became
> > inaccessible all the services were halted and we had to restart it.
> > We suspct that it was excesive paging that caused the problem.
> > How can we prevent it from happening again.
> >
> > Thank you
>
> Use one instance instead of 10.
>
> Regards
> Michel Cadot

dual Xeon 3.06 GHz 1 MB cache, 4 GB system memory.

so I was applying the 9.2.0.5 patchset to a RHEL 3.0 ES box today, to
hopefully fix an error that has been creating lots of those fun to
read 7445 .trc files.
(distributed trans over a db_link, found lots of similar issues in
metalink forums, some of which _might_ be resolved in 9205).

OUI (the 10.1.0.2 version) was running on the server console in the
server room, so I was running top over ssh from my desktop to see when
things were requiring my attention.

coming from a win32 background for much of my oracle experience, I was
surprised when I examined the amount of swap space used.

zero.
0 KB.

yeah, I hadn't turned of the filesystem buffering yet, so 3.2 GB of
memory was in use. still, not a byte of swap allocated.

I was most impressed.

I guess that one does not have to leave 50% of physical memory free,
just to appease the paging god on Linux.

One thing that I did not like in the 9205 OUI pre-install tasks was,
requiring shmmax to be 2 GB. For my purposes, 1 GB is sufficient. At
least it allowed me to check the box and proceed.

Pd

Next time, have the SA induce a system dump. Take that dump, run it through
crash(1m) and get a process listing at time of dump.

Use that listing to confirm the BIG memory/paging space users at the time.

Enable Memory Load Control so you'll not lose access to the server.

Sounds to me like your system ran out of paging space.

-Kevin

"DBA" <marcelmix@yahoo.com> wrote in message
news:3607af43.0404061304.4c4e749f@posting.google.c om...
> I have 10 oracle instances runnig on IBM AIX 5.1 P6000 server that has
> 8GB of RAM and 4GB of Swap. Last weekend the server became
> inaccessible all the services were halted and we had to restart it.
> We suspct that it was excesive paging that caused the problem.
> How can we prevent it from happening again.
>
> Thank you

Can you tell me how to enamble memory load control?
You are right the server did run out of the paging space. Would
codsider increasing paging space as the best solution?

Thank you

-Marcel



"Jeff Jackson" <nospam@nospam.com> wrote in message news:<7qWdnegPOvu_8-ndRVn_vQ@giganews.com>...
> Next time, have the SA induce a system dump. Take that dump, run it through
> crash(1m) and get a process listing at time of dump.
>
> Use that listing to confirm the BIG memory/paging space users at the time.
>
> Enable Memory Load Control so you'll not lose access to the server.
>
> Sounds to me like your system ran out of paging space.
>
> -Kevin
>
> "DBA" <marcelmix@yahoo.com> wrote in message
> news:3607af43.0404061304.4c4e749f@posting.google.c om...
> > I have 10 oracle instances runnig on IBM AIX 5.1 P6000 server that has
> > 8GB of RAM and 4GB of Swap. Last weekend the server became
> > inaccessible all the services were halted and we had to restart it.
> > We suspct that it was excesive paging that caused the problem.
> > How can we prevent it from happening again.
> >
> > Thank you