From: Webpro (aielloster@gmail.com)
Date: Thu Dec 15 2005 - 19:54:28 EST
Message from Victoria...
I have been trying to send this message to the Sunmanagers listserv
and it kicks back. Can you inform everyone with what is happening? I
had to put in a service request with SUN to get the answer. This is
what I sent to an Oracle listserv. Here it is:
I had just purchased another SUN V440 to be my backup and recovery
box. I got the system configured, modified the /etc/system file, did
the oracle install and tried to start up a database. I got the
dreaded:
ORA-27102: "out of memory"
I knew what that meant, the shared memory values were set too low. I
knew that was impossible since I have another V440 with the same
values and it works just fine. I worked on it off and on for almost 2
days. I finally went to Sunsolve. You will not like the answer.
Sun is making 'corrections' to the kernel. The IPC message, shared
memory and semaphore sections are not loaded at reboot time. They
even went so far as saying that this is a 'feature' of Solaris 10.
This is not nice. If you are considering adding any patch sets dated
11/11 or later, it removes the automatic load of these and oracle will
not open the databases. To handle this, I had to enter forceloads in
my /etc/system file before my set commands that look like this:
forceload: sys/semsys
forceload: sys/shmsys
After these statements, it worked just fine. This is also an issue
with EMC SAN. It uses semaphores. Needless to say, I was not happy.
Victoria
>>> Webpro <aielloster@gmail.com> 12/15/2005 1:33 PM >>>
We found no workaround. There is no explanation why Solaris would not
apply the /etc/system values. I think there is something more
sinister going on!
We reinstalled Solaris and were able to set the desired SHMMAX values
in /etc/system.
My guess? some system program was corrupt or zero bytes. I have seen
this before.
-- Thanks to the two who replied.
On 12/13/05, Webpro <aielloster@gmail.com> wrote:
> I am having a severe issue with shared memory on SOlaris 8. We have a
> number of machines all configured the same. Servers are in pair.
> Active and hot standby. On one pair, the standby will not obey our
> shared memory configuration. This is causing our Database to fail!
>
> The /etc/system files are identical. We configured set
> shmsys:shminfo_shmmax = 0x60000000 a number of times. with no luck
>
> We have 6GB memory on a V240 with two processors.
>
> Normally /usr/sbin/sysdef | grep SHMMAX gives this:
> 1610612736 max shared memory segment size (SHMMAX)
>
> but on the bad server, we get this no matter what we do!
> 536870912 max shared memory segment size (SHMMAX)
>
> Patch levels are the same.
>
> I would appreciate any suggestions.
>
>
> --
> "Despite the high cost of living, it still remains popular!"
>
-- "Despite the high cost of living, it still remains popular!" _______________________________________________ sunmanagers mailing list sunmanagers@sunmanagers.org http://www.sunmanagers.org/mailman/listinfo/sunmanagers -- "Despite the high cost of living, it still remains popular!" _______________________________________________ sunmanagers mailing list sunmanagers@sunmanagers.org http://www.sunmanagers.org/mailman/listinfo/sunmanagers
This archive was generated by hypermail 2.1.7 : Wed Apr 09 2008 - 23:37:49 EDT