Mem Checking Help

[Steven Chen_1]

I would appreciate it very much if someone can give me some confidence about how to check memory on 10.20 of K box.

I have problem on database crash with possible cause of memory (physical) or user process overflow (according to Oracle). While none is identified, i.e. no message indicates OS problem and memory bad, as well as database processes bad, I would see if hp guru have any thought.

dmesg, ipcs, glance, swapinfo are what I have used.

Thanks.

Steven

[James R. Ferguson · Accepted solution]

Hi Steven:

From a software (O/S) perspective, you are using the right tools to assess memory utilization.

IF you want to satisfy yourself that your physical memory is good you can view its configuration and error levels with the Support Tools Manager (STM):

# echo "selclass qualifier memory;info;wait;infolog"|cstm > /tmp/meminfo

Remember that single-bit errors are automatically corrected, and should not impact application software. A growing number of single bit errors is a warning to get a memory sim replaced, however.

...JRF...

[Ravi_8]

Hi,
SAM-->performance monitor-->system properties-->meory gives the physical memory and swap info

If the database showing process over flow check the nproc in kernel parameter, orcale recommends 4096.

[linuxfan]

Hi Steven,

To determine the actual physical memory available on the system.

1. dmesg | grep Physical
(if it hasn't been too long since the machine reboot).

2. /usr/sam/lbin/getmem
(This command has to be run by root, gives the output in MB)

If you are looking for the actual location of the symm and their sizes, one of the easiest ways without reboot the system is to use cstm/mstm/xsm.


-HTH
I am RU

[JACQUET]

Hi,

You can use this C-program which shows you the available memory of you K server.

PJA

[Steven Chen_1]

Thank you all for the great help. I am unable to assign points yet.

For deeper inspection, can you help finding any problem on ninode in kernel parameters? The sar -v result is:

-----------------------

#sar -v 5 5

HP-UX B.10.20 C 9000/879 08/14/01

10:08:50 text-sz ov proc-sz ov inod-sz ov file-sz ov
10:08:55 N/A N/A 197/3620 0 3894/4418 0 814/6779 0
10:09:00 N/A N/A 197/3620 0 3893/4418 0 814/6779 0
10:09:05 N/A N/A 197/3620 0 3895/4418 0 814/6779 0
10:09:10 N/A N/A 197/3620 0 3888/4418 0 814/6779 0
10:09:15 N/A N/A 198/3620 0 3894/4418 0 817/6779 0

---------------------

I see inod-sz is close. Do I have to adjust it from kernel?

Thanks a lot.

Steven

[linuxfan]

Hi Steven,

You can change the kernel parameters is multiple ways. One of the ways is use SAM to change the kernel parameter. Mind you this change will require a reboot of the system.

To find out more information about the various configurable kernel parameters you can look at
http://docs.hp.com/hpux/onlinedocs/os/KCparams.OverviewAll.html

-HTH
I am RU

[Bill Hassell]

It is very unlikely that a true RAM problem such as a parity error) would go unnoticed...there will be lots of logged errors. It's unfortunate that the actual reason for the Oracle problem does not return a Unix error code as this would greatly simplify the fix.

The sar -v output you are looking at has only 2 useful values, proc-sz and file-sz. The inode-sz entry is a measure of the cache usage and should normally be 100% after a few hours following a reboot.

The most common parameters that need changing for database servers are:

nproc
nfile
maxdsiz
maxuprc

There are application-specific values such as semaphores but these are usually very explicit from the database manufacturer. Start by doubling nproc, nfile and maxdsiz. maxuprc won't need to be changed unless you have dozens of users logging in as the same username.

[Praveen Bezawada]

Hi
When running oracle it helps to check the values of
dynamic cache
dbc_max_pct

Shared memory related values
shmmax
shmmni
shmseg

also check the value of semaphores
SEMMNS
SEMMNI

...BPK...