HP NonStop G-Series - Manual (Page 60)

5.4.3. Alert Considerations
These are things to consider before creating alerts.
 All alert names must have unique names meaning that you cannot create an
alert with the same name as any other alert, pre-defined or user-defined.
 There are pre-defined alerts set for CPU Busy counter and Process Busy
counter that can be edited, but not deleted. These pre-defined alerts for these
two counters are named as CPU CRITICAL, CPU WARNING, PROCESS
WARNING and PROCESS CRITICAL.
 The scope of the alert is dependent on the level in the exploration tree at
which it is set. For example, if you set an alert at the node level, the alert will
only be for that particular node. And likewise, if you set an alert at higher
levels in the exploration tree, the alert will apply to all that fall under that
scope.
 Alerts for disk should not be created at higher levels if you are running both
Linux and NonStop agents because this will cause the alerts not to be enacted.
Refer to Section
Alert Logging for additional information.
 Thresholds created at the higher levels will not seamlessly apply to
nodes added into the cluster after the creation of the threshold.
 For example, a threshold Process – Busy – Greater Than 60 – Minor
(PrBsyGT60Min) has been set at the NonStop Cluster level for a cluster
containing nodes NS1 and NS2. A node NS3 is added to the NS cluster
after this threshold is set. If a process goes above 60% on either NS1
or NS2, the status would change to Minor and the alert will be logged.
However if a process goes above 60% on NS3, no effect takes place.
 To set the threshold for NS3, create a new threshold (with a different
name) at the Node level for NS3 or delete the existing threshold
(PrBsyGT60Min) at the higher level and recreate it at the same level.
[Back to TOC]
Performance Essentials HP SIM Client and NonStop & Linux Host User Guide – 544813 – 004
60