Owners Manual
Understanding Performance Monitoring | Performance Monitoring
OMNM 6.5.2 User Guide 363
The following items correspond to the numbered processes in
diagram 1
.
Process Description
1
For each enabled
Monitor and each
target
This process executes for each item within the Resource Monitors portlet that
is marked as Enabled and for each targeted entity.
2
Polling interval has
elapsed
This process executes every time the Polling Interval has elapsed. For example
if the Polling Interval for a given Monitor is 15 Minutes then it executes that
often.
3
Server low on memory
Is the server low on memory? This is determined by the property
lower.threshold.limit
which by default is 5. This means that if the
available memory on the server is less than 5% (or whatever value this property
is set to) then it meets this condition.
4
Skip due to low
memory
Numbers 4, 7 and 8 all involve skipping rather than trying to poll the targeted
entity. If OpenManage Network Manager finds the system low on memory
then it cannot poll the target(s) and skips polling instead. Any time it skips it
increments the associated skip count and also updates the associated last time
skipped. You can view these attributes can be viewed through the JMX console
> oware > service=PollingEngine. See
SkippedExecutionDueToMemoryCount
and
MostRecentSkippedExecutionDueToMemory
.
5
Is the server done
processing the
previous polling
attempt for all targets?
Is the server still working on polling targets from the previous time? The
polling process begins every time the polling interval elapses, regardless of
whether or not the server is still working on the previous round of polling.
This means the previous polling attempt has not produced results so the
server will have to skip it (due to overload) rather than trying to perform two
(or more) polling attempts simultaneously. If the polling interval is short
enough then there will inevitably be skips due to overload because the polling
process takes time.
6
Is the polling thread
hung?
Is the polling thread hung? If so it is not doing any work and will not produce
any results. There is a maximum duration that a polling thread has to produce
results before the polling engine considers a thread hung. The PollingEngine
MBean attribute
ThreadInterruptThreshold
which is set to 90000
(15 minutes) by default determines this.
7
Skip due to hung
thread
Similar to number 4. If a polling thread is hung then it cannot poll the
target(s) and will instead skip. Any time it skips it will increment the
associated skip count and also update the associated last time skipped. View
these attributes through the JMX console > oware > service=PollingEngine.
See
HungThreadInterruptedCount
and
MostRecentHungThreadInterrupted
. When the system identifies a
hung polling thread it tries to interrupt the thread and reclaim it.
8
Skip due to overload
Like number 4. If the server is overloaded then it cannot poll the target(s) and
will instead skip. Any time it skips it will increment the associated skip count
and also update the associated last time skipped. View these attributes
through the JMX console > oware > service=PollingEngine. See
SkippedExecutionDueToOverloadCount
and
MostRecentSkippedExecutionDueToOverload
.
9
For each attribute
Every Monitor has a set of attributes that it will try to periodically poll from
each targeted entities.
10
Attempt to poll this
attribute
OpenManage Network Manager tries to reach the target device and poll the
given attributes.