What is the important parameter ?

[insik,Kang]

Dear everyone.

I had terrible experience. Last Week, my server (HP UX 11) had some problem.
One of the parameter of that server, nfile was 30000.
When my users starting working, It was over.
The Oracle was down immeately.
HP engineer said to me It occured because of nfile parameter. So It is now 50000.

I just want to know What is the important parameter that i have to mornitoring?

I use the "sar -v" to check the nfile parameter .
But there are 180 parameter in my server.
Plz let me know Which parameter is important?
which one I have to check and how to check?

Best Regards..

[Michael Tully]

There are quite a number of important parameters. These include nproc, nfile, shared memory, dbc_max_pct and various others. The best source document that explains these can be found here:

http://docs.hp.com/hpux/onlinedocs/939/KCParms/KCparams.OverviewAll.html

If you have the 'glance' tool installed. (/opt/perf/bin/glance) it gives a better overview than most. (sar -v included) You can install a trial license version from your application CD set.

Some of these are also have values in formulas instead of values. One of the best places to see these is in 'sam'.

[RAC_1]

nfile is systemwide no. of open files.

The usage of which can be monitored with sar -v and glane -t.

Depending on this usage you can set it to reasonable value.

We have set it to 30000 and we harly hit 10000

Please check what exactly caused oracle shutdown. Did you get any error message?

[Bhuvaneswari Selvaraj]

The parameter that you have to modify is the nfile parameter, You can get the value of the smae by using the kctune command

kctune nfile

will tell you the current value.

kctune nfile=50000

will change its value.

You can use the "Event Monitoring Services" Application, that sits under "Resource Management" in SAM to monitor /system/kernel_resource/file_system/nfile and to notify you when it meets a thershold (you can define the threshold) It notifies you by any notification method that you specify,like email, snmp and so on...

[Bhuvaneswari Selvaraj]

Once you bring up the EMS GUI, under the "Action" menu, choose "Add monitoring request" . The rest you can do it yourself, it is a somple GUI.

Also, you can use resls to find the current value of nfile

resls -s /system/kernel_resource/file_system/nfile

Among all the parameters, we monitor nfile, nproc and maxfiles

resls -s /system/kernel_resource/process_management/nproc