
Chapter 8
Statistics and Monitoring
8.1 Introduction
The Rhino SLEE provides monitoring facilities for capturing statistical performance data about the cluster using the client side
application
rhino-stats
.
To launch the client and connect to the Rhino SLEE, execute the following command:
>cd $RHINO_HOME
>client/bin/rhino-stats
One (and only one) of -g (Start GUI), -m (Monitor Parameter Set), -l (List Available Parameter Sets) required.
Available command line format:
-d : display actual value in addition to deltas for counter stats
-R : display raw timestamps (console mode only)
-w <argument> : password
-h <argument> : hostname
-H <argument> : bind address for direct statistics download
-C : use comma separated output format (console mode only)
-p <argument> : port
-P <argument> : port for direct statistics download
-g : gui mode
-l <argument> : query available statistics parameter sets
-q : quiet mode - suppresses informational messages
-i <argument> : internal polling period in milliseconds
-m <argument> : monitor a statistics parameter set on the console
-u <argument> : username
-f <argument> : full path name of a saved graph configuration .xml file to redisplay
-t <argument> : runtime in seconds (console mode only)
-T : disable display of timestamps (console mode only)
-j : use JMX remote option for statistics download in place of direct statistics download
-n <argument> : name a tab for display of subsequent graph configuration files
-s <argument> : sample period in milliseconds
The
rhino-stats
application connects to the Rhino SLEE via JMX and samples requested statistics in real-time. Extracted
statistics can be displayed in tabular text form on the console or graphed on a GUI using various graphing modes.
A set of related statistics is defined as a parameter set. Many of the available parameter sets are organised in an hierarchical
fashion. Child parameter sets representing related statistics from a particular source contribute to parent parameter sets that
summarise statistics from a group of sources.
One example is the
Events
parameter set which summarises event statistics from each Resource Adaptor entity. In turn each
Resource Adaptor entity parameter set summarises statistics from each event type it produces. This allows the user examining
the performance of an application to drill down and analyse statistics on a per event basis.
Much of the statistical information gathered is useful to both service developers and administrators. Service developers can use
performance data such as event processing time statistics to evaluate the impact of SBB code changes on overall performance.
For the administrator, statistics are valuable when evaluating settings for tunable performance parameters. The following types
of statistics are helpful in determining appropriate configuration parameters:
Three types of statistic are collected:
• Counters count the number of occurrences of a particular event or occurrence such as a lock wait or a rejected event.
56
Kommentare zu diesen Handbüchern