Analyzing data with dim_STAT

dim_STAT is a tool used for monitoring and performing a detailed performance analysis of Solaris/Linux systems .

Overview of dim_STAT :

Let us suppose a number of Solaris and Linux machines need to be monitored in a multi host environment. dim_STAT is a good utility which could be used in this situation.

dim_STAT contains various components like MySQL database, Webserver, WebX tool, STAT-service etc. The working of the various components are described below -

The STAT-service daemon(client) need to run on all the hosts. This daemon will gather data by running various services like vmstat, iostat and netstat on the hosts. Collectors could be started via a Web browser to communicate with the clients.

These collectors will then feed the collected data into the MySQL database. This data will then be analyzed using the WebX tool. The WebX tool can give its analysis in various forms like table, text, graphs etc.

The WebX tool is actually a CGI which is hosted on top of a webserver. The MySQL database, the webserver and the WebX will be installed during the installation of dim_STAT. Further, the MySQL and the webserver will be started when dim_STAT is started.

The data loaded in the MySQL could either be the data collected during an online monitoring or could be the data which was previously collected. Previously collected data could be stored in the form of a tar.Z file and can be used later for data analysis.

dim_STAT installation

The latest dim_STAT could be downloaded from (dim_STAT.tar)

After an untar, go into the dim_STAT-INSTALL directory and run as the root. Enter the various parameters asked. These parameters are used for installing WebServer and MySQL. The default parameters should work.

The /apps is the default dim_STAT home directory and /apps/ADMIN contains all the administration scripts.

You can then start the dim_STAT by using the following commands -

# cd /apps/ADMIN 
# dim_STAT-Server start

To stop servers:

# cd /apps/ADMIN 
# dim_STAT-Server stop
Loading data

Just after installation, dim_STAT will provide a link in the output to access the “Home Page” of dim_STAT.

The “Home Page” looks like this -

Now, click on the “Welcome” link. Note that a link to a complete user guide is given just below the “Welcome” link.

The “Welcome Page” looks like this -

If you want to do online monitoring of the data, click on the “Start New Collect” link. Enter the information to start a new collect on the host named localhost. Click the “Continue” button to move to the next screen. Select the monitoring options of your choice (for example: vmstat, iostat, netstat). Finally click the “Start STAT(s) Collect Now!!!” button to start monitoring.

If you need to analyze a previously collected data, in the “Welcome Page”, click on “Restore Collects from tar.Z”.

Upload the tar.Z file and click on “Restore”. dim_STAT will unzip/untar the file supplied. It will also logically group the files according to the various runs and display them. The runs that need to be analyzed should be chosen and restored. This will load the corresponding data into MySQL.

Changing the database

By default, the database created for storing the data is dim. If you want to create another database for storing the data, click on “Preferences” link on the “Welcome Page”.

In this page, you can create a new database and set any existing database as the default.

The current (working) database name is always present in every dim_STAT window as “Database : [db-name]”.

Analyzing the data:

To start analyzing the data loaded, click on the “Analyze” button on the “Welcome Page”.

You could either do a “Multi-Host Analyze” or a “Single Host Analyze”. The single host is described below

Select “Single-Host Analyze” and click on the “Analyze” button to proceed to the next screen.

Choose the log file you want to analyze and click on the monitoring option of your choice. Eg : let us suppose we click on psSTAT.

Click on the corresponding user and the process name. The “Interval” section gives a choice to select the exact time period that we need to be monitored. In the “Graphics” section, you can choose if you want an applet or a PNG image as the output. The background color and the size of the image could also be modified as needed.

Then, click on the various values that need to be analyzed (eg : CPU%, MEM%) and click on “Start”.

A sample graph is given below -


BTW, you may also enable jvmSTAT Add-On and collect JVM-specific data from all your running java processes! :-)

Posted by add on June 26, 2008 at 11:21 AM IST #

Post a Comment:
  • HTML Syntax: NOT allowed



« June 2016