This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
The system logs are wh ere the operating system an d som e app lications write event records.
Depend ing on the operating system, there may be mu ltiple logs.
:K\
There may be indications of a developing p roblem (for examp le, a hard dr ive that generates
errors may indicate that it is failing and needs to be replaced).
+RZ
*XLGHG7RXU
1. In the Command field, enter transaction OS06 and choose Enter (or from th e SA P standard menu, choose Tools → CCMS → Control/Monitoring→ Performance menu →
The uninterrup tible power supp ly (UPS) that you u se shou ld be monitored by a control
program. This program, wh en triggered by a pow er event, records the event an d initiates a
shutd own p rocess of the R/ 3 environmen t (R/ 3, the database, related ap plications, and th e
operating system), and finally the server. In ad dition, most UPSs have a self-test and
capacity calibration fun ction. The resu lts of these tests are logged . Specific data logged
depend s on the program and the UPS.
:K\
You n eed to review the p ower events tha t triggered the UPS control program .
While the UPS protects the server, the control program should be recording pow er eventssuch as pow er dips, brown outs, power failures, etc. This record ing could help you or the
facilities person solve electrical problems in th e facility. For examp le, a pattern of pow er
dips or ou tages may indicate a problem elsewh ere in the building.
You need to verify that:
< The UPS is functionin g
< The self-tests comp leted su ccessfully
< There is sufficient capacity in the ba tteries
The batteries in the UPS mu st be period ically replaced. If the batter ies are low, the cap acity
test will indicate that the batteries do n ot have sufficient capacity to shu tdow n the system
before failing.
+RZ
Review th e log for the UPS control program .
&KHFN\RXU8366KXWGRZQ3URFHVV
:KDW
Verify that your UPS shutd own process works. A shu tdow n pr ocess is an autom ated script
for the UPS to shut d own R/ 3, the database, other app lications, the operating system, andthe UPS.
:K\
This check verifies that the entire shu tdow n process works as planned an d d ocumented .
When there is a power failure, the R/ 3 environment should be shu t dow n in an orderly
man ner. There shou ld be sufficient reserve in the UPS to reach the en d of the shu tdow n
process. Something m ight have changed since your last test to cause the shu tdow n p rocess
to fail. If this process fails, you n eed to find out wh y an d fix the p roblem.
The stopsap comman d d oes not work w ithin all UPS control programs. You need to
verify that you r UPS control program will properly stop R/ 3 and the database before
shutting d own the server.
Like a car battery, UPS batteries wear ou t over time an d mu st be replaced. If the battery is
worn out, the UPS will not hav e sufficient power to complete the shutd own process.
&KHFN0DLQWHQDQFH&RQWUDFWV
:KDW
Many of the servers and related equ ipmen t are und er maintenan ce or service contracts with
the manu facturer or distributor.
< The produ ction system and critical equipment shou ld be und er a “prem ium” 24 hour x 7
day (x 2 hou r response) sup port agreemen t.< Less critical equipm ent can be un der a next-business-day support agreement.
:K\
If you n eed sup port or service and the service contract has expired, the confusion an d time
to reestablish the service contract could be critical.
:KDW
The supp ort level should be selected based on equ ipment u se. If a piece of equipm ent
becomes critical to the compan y’s operation, its supp ort level should be u pgraded to reflect
the critical nature of tha t equipment. Conversely, equipment could become “noncritical” orbe replaced. In this situation, the service contracts could be d own graded or d ropp ed as
appropriate.
+RZ
< Keep a list of service contracts.
Includ e wh at these contracts are for and the expiration d ate in the list.
< Review equipm ent usage to determine if the supp ort level for equipm ent should be
upgraded, downgraded, or dropped.
< Review the list for expiration d ates each quarter.
How long in advance of the expiration date to do this review dep ends on the time ittakes to go through th e pu rchase requisition and ap prova l process in your comp any.
This example, however, has not been configured to p ick up and report on SAP events.
Initially, there will be a lot of tuning as th e system p aram eters are adjusted. Over time,
parameter adjustments will redu ce.
:K\
You m ay need to change alert parameters to filter noncritical events and to generate alerts
for critical even ts. The key to rem ember is that this process is dyn amic. Some of these tasks
are as follows:
< Accoun t for new events that have never occurred .
Critical, you need to generate a p age
Importan t, you need to generate a m essage (for examp le, e-mail)
< Determine if an event th at used to be filtered now needs to generate an alert
< Filter out events (both old an d n ew) that shou ld not genera te alert messages
Filtering is necessary to m anage th e messages that are r eviewed. If too man y irrelevantmessages get through the filter, it becomes d ifficult to review the alert m essage log.
< Adjust for personnel changes
There may be oth er events that r equire action (for examp le, shift or d uty chan ges for
organizations with several people “on call”).
< Test that all alert m echanisms are fun ctional.
The paging/ messaging fun ction needs to be tested regu larly. If the monitoring program is
unable to send a page, you will not receive the page w hen a critical alert occurs.
The inability to send a page can be caused by:
< Someone changing something in the e-mail or phone system th at preven ts alertmessages from being sent.
< A phone patch cable that has disconnected from the modem.
+RZ
To review the paging system:
< Review the var ious monitored logs (such as the NT event logs) to look for events that
should generate an alert message (e-mail or page).
The monitor program needs to be configured to pick these events up and prop erly
process them.
< Review the a lert monitor log for alert events that should be “filtered” out.
The monitor program needs to be configured to filter or ignore such even ts.
< Test all alert mechanisms, such as p ager, e-mail, etc. to make sure th at they are
functional.
If you receive regu lar d aily e-mail messages, then the e-mail testing is being done for