The Cluster Summaries MonALISA Module

Introduction
Installation of the Module
Accessing the Cluster Summaries

Introduction

The Ganglia Monitoring Tool provides average metrics per cluster. If you have installed version 2.5.4 or higher, you can connect to port 8652 on the host that runs the gmetad deamon and send the string:

            /?filter=summary

You get an XML file with averages about all clusters. In addition you may just request averages about a single cluster by simply passing the name of the Cluster:

           /STAR CAS Linux Cluster

If you pass the name of a host after the Cluster you get the metrics for that host:

           /STAR CAS Linux Cluster/rcas6156

We developed a MonALISA monitoring Module that pubishes the average metrics per cluster. The module opens a Socket to port 8652 to the host that runs the gmetad deamon and writes the string /?filter=summary to the Socket. Currently it only publishes the total number of CPUs (NCPUS) per cluster, the sum of load1 for all compute nodes in each cluster (Load1), the sum of load5 for all compute nodes in each cluster (Load5), the average load1 (Average1) which is calculated as Load1/NCPUS and the average load5 (Average5) which is calculated as load5/NCPUS. The module can of course be extended to publish averages of other metrics that are available in Ganglia.

Installation of the module

Make sure the host where you installed the module is in the trusted_host field in the ganglia configuration file. Try the following:
```
                      telnet hostname 8652
                  
```
You should get the expected XML file.

Download the tar file with the ML module myGanglia.tar and untar it under the usr_code directory:

                     cd $MonaLisa_HOME/Service/usr_code
                     tar -xvf myGanglia.tar

A new directory, under usr_code, will be created (myGanglia/) that contains the ML module (aGanglia2), a script comp to recompile the module if you make changes to it and a script test_run to run the module from the command line before implementing it with the ML monitoring Service.
As with any other ML custom module, you must include the directory where you installed the module ($MonaLisa_HOME/usr_code/myGanglia/) in the lia.Monitor.CLASSURLs property. This property is set in the ml.properties file. If you are using other custom modules in other directories you must separate them with commas. Do not forget the "/" at the end of the directory.
```
lia.Monitor.CLASSURLs=file:${MonaLisa_HOME}/Service/usr_code/LSF/,file:${MonaLisa_HOME}/Service/usr_code/myGanglia/
                  
```
Edit the <farm_name>.conf config file and add a new cluster:
```
    *CLUSTERS{aGanglia2, localhost, ganglia01.rcf.bnl.gov,8652}%30
             
```
(change the hostname and the port number above).

Restart MonALISA:

                   cd $MonaLisa_HOME/Service/CMD
                   ./ML_SER restart

Accessing the Cluster Summaries

Since we run the MonALISA service as a web service you may access the information we publish in MonALISA using the Web Service Clients. You may also access the same informaton and make plots using the Global MonALISA client.

Stratos Efstathiadis - page was last modified