Red Hat Bugzilla – Bug 819741
All metrics under an ESB service's "Monitoring" tab don't seem to be suitable for a monitoring purpose in JON 3.0.
Last modified: 2012-05-08 02:43:13 EDT
Description of problem:
On JON 3.0, all metrics under an ESB service's "Monitoring" tab seem to be based on cumulative data and don't seem to be suitable for a monitoring purpose. "Message Count" in ESB metrics should represent a momentary value than a cumulative one.
Version-Release number of selected component (if applicable):
JON 3.0.GA / SOA-P 5.2.0
Steps to Reproduce:
1) Install and set up a SOA-P instance.
2) Start the SOA-P instance.
3) Add a new SOA-P instance to JON inventory - JON server/agent must have SOA-P 5 plug-in installed.
4) Execute $SOA-P_HOME/samples/quickstarts/helloworld sample.
4.1) run "ant deploy"
4.2) run "ant runtest"
4.3) wait 20min+.
5) Repeat the steps 4.2) and 4.3) several times. In 4.3) you need 20min+ as an interval because JON collects metrics every 20min by default. You could set a shorter interval on JON.
6) From left navigation tree, expand and select a node in JON.
7) From right pain, select "Monitoring" tab -> "Tables" and see metrics.
8) From right pain, select "Monitoring" tab -> "Graphs" and see metrics.
- In "Tables", a value of "Message Count" row and "Maximum" column seems to show a cumulative value. See "monitoring-tables.png" I attached to this case. In this example, the value was 3 as I repeated the steps 4.2) and 4.3) three times.
- In "Graphs", value of "Message Count" graph doesn't seem to go down and also seem to show a cumulative value. See "monitoring-graphs.png" I attached to this case. In this example, the value was 3 as I repeated the steps 4.2) and 4.3) three times.
With that, "Maximum", "Average" and "Last" columns seem to show non-suitable values for a monitoring purpose because they are derived from the cumulative values.
I expected "Message Count" in ESB service's metrics should represent a momentary value than a cumulative one. Otherwise statistic metrics based on the "Message Count" metric would not correctly represent monitoring metrics IMO.
For an example, metrics under an EAP node such as memory usage is a typical momentary value and the statistic data based on that metrics would be suitable for a monitoring purpose. However, "Message Count (avg)" in ESB service's metrics seems to represent an average based on a cumulative metric("Message Count") and it's would not be suitable for a monitoring purpose. Assuming JON's data collection interval is 20min and a time A is the first time when JON collects ESB's service data, say
at time A, "Message Count" is 100,
at time A + 020min, "Message Count" is 200,
at time A + 040min, "Message Count" is 300,
at time A + 060min, "Message Count" is 400,
at time A + 090min, "Message Count" is 500,
then the latest "Message Count (avg)" would be 300 but it would not be useful for a monitoring purpose as it's based on cumulative data.
Is it a know bug? Or am I missing something?
└ JBossAS Servers
└ SOA-P localhost:1099 default
└ JBoss ESB
└ SimpleListener Service
Note: I referred to the ESB administration guide . I attached screenshots of "Graphs" and "Tables"
Created attachment 582865 [details]
screenshot of montoring - tables
Created attachment 582866 [details]
a screenshot of montoring - graphs