Is it possible to get an alert when database page faults exceeds a certain amount.
Is it possible to get an alert when database page faults exceeds a certain amount. Iam referring to the page faults which appear when we give the WRKSYSSTS command. Or write a pgm which could monitor the page faults and sent an alert.
Software/Hardware used:
iseries system i WRKSYSSTS
Software/Hardware used:
iseries system i WRKSYSSTS
ASKED:
January 12, 2010 12:59 PM
UPDATED:
January 18, 2010 6:52 AM
Answer Wiki:
Use iSeries Nav to create a system monitor. Under the Metrics tab, add Machine Pool Faults, set your threshold and send to your email and/or cell phone.
==============================================================
You have at least a couple options.
Probably the easiest is the <a href="http://publib.boulder.ibm.com/infocenter/iseries/v5r4/index.jsp?topic=/apis/qwcrssts.htm">Retrieve System Status (QWCRSSTS) API</a> using format SSTS0300 for database faults/pages. You can choose "Reset status statistics" *YES or *NO depending on whether you want statistics snapshots or cumulative. Obviously, you can create any kind of alerts that you want.
More difficult (and less well known) would be the <a href="http://publib.boulder.ibm.com/infocenter/iseries/v5r4/index.jsp?topic=/apis/perfmgmt.htm">Performance Management APIs</a> which you might connect with collectors/monitors that you define in Management Central. For example, the <a href="http://publib.boulder.ibm.com/infocenter/iseries/v5r4/index.jsp?topic=/apis/qypsacno.htm">Add Collector Notification (QypsAddCollectorNotification) API</a> "registers with a collector to provide notifications to a specified data queue for a collection event." (You're on your own as far as exploring these APIs goes, but a lot of them seem appropriate.)
Personally, I'd start with QWCRSSTS if I was looking for a simple in-house monitor. I've used it to good effect for years and simply haven't needed more.
Tom
To see all answers submitted to the Answer Wiki: View Answer History.
Thanks its working. However this monitor will stop the moment i exit the Navigator. Is there any way by which this can continue and trigger raised whenever the trigger point is reached.
We use iNav in a similar manner. The i5/OS trigger command on a few of our message monitors use SNDDST to send notification to a cell phone. I can confirm that message monitors stays alive even after iNav is ended on your PC because I received several messages over the weekend.
Since message monitors stay alive, that leads me to believe the system monitors may also stay alive. While it won’t fire the iNav status screen it may execute the trigger command.
iNav is a GUI for the iSeries so there is a companion job started if you are using Management Central. Therefore if you exit iNav, or for that matter don’t start it the next day, the companion job will still be running on the system. From the green screen, enter wrkjob qypsjsvr and view the joblog. You will see your monitor(s) there.
You can also view the qypsjsvr job from iNav. Expand My Connections, your system, then Work Management. Click on Server Jobs. Under the Server column, look for Management Central Server. Right click on select joblog to view. If it is not running, enter STRTCPSVR SERVER(*MGTC).
Is it possible to change an existing monitor which is running, eg change the trigger command, interval etc. How do we do it.
I assume the monitor will continue to run till the next ipl. Will the monitor job start automatically on the next ipl.
Also will it create any extra files and where, while this job is being run. Can these be deleted or do they automatically get deleted. Do these take huge disk space.?
Thanks for the tip, Whatis23. I had never looked into the job log for QYPSJSVR.
You can change your monitor while it is active but it will not take effect until you end then restart.
Right click on the monitor, select stop then start afterwards.
It will restart after an IPL but you can add STRTCPSVR SERVER(*MGTC) to your startup program.
As far as i know, no files are generated or retained.
Once the trigger is alerted we change the max active threads in the sharepool by using the formula of totalling all the jobs that are currently running in the sharepool
(this total is arrived by using WRKSBS command to find the currently running jobs in all the active subsystems-which is attached to each share pool)and adding 10% more jobs to the currently active number of jobs in that pool, i hope you have understood .
Is there any way by which this process can be automated ie changing the max active threads once the threshold is reached.?
Can we call an API