Sudden Reboot

[CowBoy]

Dear All
My OS(HP-UX) will reboot immediately laft no log in log directories.
I have rp8420+hp-ux+Oracle 10g
The only change that i have made was increasing the shmmax equally to the ram(65 GB)
The most confusing thing is that there is no log in any directory!!!!!!!
Thank you in advance

[Hasan Atasoy]

hi;


any log in the console logs ?


Hasan.

[CowBoy]

Unfortunately no Hasan
I am getting mad!!!!!!!
I have never had somthing like this....

[Mark McDonald_2]

Can you rule out a power supply issue? Are you using UPS and redundant power supplies?

It may be a hardware problem? The system panic'd. I would try looking closer at the console logs.

[Ninad_1]

Hi,

Has it happened just once or it is happening often ?
Did you check the console logs ?
Did the system crash or rebooted ? Did you check your crashdump area ?


Regards,
Ninad

[CowBoy]

Dear All
I think i am confusing you friends!
after i decrease the SHMMAX everything goes OK.
and know everythiong is normal but i can not find the relation between the increasing the SHMMAX and sudden reboot???

[Ninad_1 · Accepted solution]

Hi,

Just increasing the SHMMAX parameter as such should not mean anything, untill some application tries to use a large shared memory segment.
SHMMAX will only define what is the contiguous maximum memory segment that can be allocated to an application. Thus say even if you have SHMMAX as a lesser value and your application requests higher shared memory segment, then there will be multiple shared memory segments that will be allocated to the application (provided there is enough virtual memory - memory+swap - available).
Thus I believe that the application request of shared memory is more important than just the parameter.
In short, if your Oracle DB is not asking for more shared memory, then just increase or decrease of SHMMAX should not affect the server operations. i.e. If your DB does not demand for more SGA/PGA, then I do not see why changing SHMMAX should affect anything - like reboot of server.

Are any other applications running other than Oracle DB ?
How much SGA/PGA is being used by the Oracle DB ? - ipcs -mb
Are you starting multiple instances of DB ?
Does the reboot happen when you start some application/DB ?
Are there any core files generated ?

Regards,
Ninad

[Dennis Handly]

(Note: Your real problem is a panic, not a "Sudden Reboot".)

>The only change that I have made was increasing the shmmax equally to the ram(65 GB)

It won't help to make it that big, you should leave room for the kernel, disk cache, etc.
What does "swapinfo -tam" show?
What OS version do you have?

[CowBoy]

Are any other applications running other than Oracle DB ?
I have One One Oracle Instance but there is a middleware which is sendding requests to the oracle DB.
We just increase the SGA_MAX_SIZE lower than half of The total memory(SGA+PGA=27G)
The total memory is 64GB.
How much SGA/PGA is being used by the Oracle DB ? SGA=17GB PGA=10
Are you starting multiple instances of DB ?
No There is Just One Instance
Does the reboot happen when you start some application/DB ?yes two min after i start my Middleware.
Are there any core files generated ?
Unfortunatly there is no log or core dump neither in OS nor in Oracle!!!!!!
PS:when there is no request toward the database Everything is fine but when i start the Middleware oracle and OS will fail without any log....

[CowBoy]

OS:hp-ux 11.23
oracle 10.2.0.1
nothing wrong in swapinfo

[Dennis Handly]

>nothing wrong in swapinfo

I asked for the output so we could check the numbers.

[F Verschuren]

Hi,
strange problem,
panic's will normaly show up in:
tail /etc/shutdown.log

What logging dit you checked (just to make sure you dint missed oneâ ¦)

If there is noting here is sounds like a environment problem and a coincident that you changed the parameter.

how have you changed the parameter?
Do you know if the parameter is not exigently decreased

[CowBoy]

swapinfo -tam just 21% of total RAM is used

[CowBoy]

hing in /etc/shutdown.log

[Dennis Handly]

>swapinfo -tam just 21% of total RAM is used

Again, I would like to see the whole output.

[CowBoy]

Mb Mb Mb PCT START/ Mb
TYPE AVAIL USED FREE USED LIMIT RESERVE PRI NAME
dev 32768 0 32768 0% 0 - 1 /dev/vg00/lvol2
reserve - 30154 -30154
memory 65485 9270 56215 14%
total 98253 39424 58829 40% - 0 -

[Dennis Handly]

>reserve - 30154 -30154

You have reserved nearly all of device swap.
Do you have Oracle running now?
What was the old value of shmmax, that worked?

As Ninad said, have you looked in /var/adm/crash/? Have you enabled crash dumps?

Are you up to date on patches?

[CowBoy]

How can i check that the crash is enabled?
oracle is now running!!!

[Dennis Handly]

>How can I check that the crash is enabled?

Check in /etc/rc.log.
There should be a file in /etc/rc.config.d/ that enables is.

[CowBoy]

It was started

[CowBoy]

close