Cluster Performance Monitoring with Ganglia and NetLogger

Salil Kulkarni, Venkatram Vishwanath, Luc Renambot

Version: March 11, 2005

ResearchGear ID: 20050311_kulkarni

This tool collects Ganglia metrics at regular interval and for a specified duration, and store the results in a NetLogger file. It will be combined with application performance metrics next. Each run will also be stored into a database for further inspection and analysis.

  • To compile on EVL cluster, just type "make":
    • On other machines, need the libxml2 compiler flags. These can be found by typing
      "$ xml2-config --libs and xml2-config --cflags" at the command line, and then compile using g++.

  • The program reads from a file called "CONFIG", the metrics that need to be recorded. The format of this file is:
    • <ganglia_metric_name> = ON/OFF on a single line where <ganglia_metric_name> is the name as it appears in ganglia's output.
    • Metrics:
      • cpu_nice = ON
      • cpu_user = ON
      • mem_total = ON
      • mem_free = ON
      • proc_total = ON
      • proc_run = ON
      • pkts_in = ON
      • mem_buffers = ON
      • bytes_in = ON
      • mtu = ON
      • mem_shared = ON
      • cpu_aidle = ON
      • cpu_speed = ON
      • cpu_idle = ON
      • mem_cached = ON
      • cpu_num = ON
      • bytes_out = ON
      • cpu_system = ON
      • pkts_out = ON

  • To run the program type:
    • "$ ./nlganglia <host_name> <Log file name> <Running_Length_in_sec> <interval_in_sec>"
    • ./nlganglia localhost out.netl 10 1

  • The program will append the NetLogger format file to a file by the name <Log file name>
  • Ouput example: 
    • t DATE: 2005-03-11T17:58:01.415647
      s LVL:  Info
      s HOST: yorda2-10.evl.uic.edu
      s METRIC: cpu_nice
      d VALUE: 0.0
      s PRG:  ganglia

      t DATE: 2005-03-11T17:58:01.415647
      s LVL:  Info
      s HOST: yorda14-10.evl.uic.edu
      s METRIC: cpu_nice
      d VALUE: 0.0
      s PRG:  ganglia

      t DATE: 2005-03-11T17:58:01.415647
      s LVL:  Info
      s HOST: yorda1-10.evl.uic.edu
      s METRIC: cpu_nice
      d VALUE: 0.0
      s PRG:  ganglia

      t DATE: 2005-03-11T17:58:01.415647
      s LVL:  Info
      s HOST: yorda13-10.evl.uic.edu
      s METRIC: cpu_nice
      d VALUE: 0.0
      s PRG:  ganglia

      t DATE: 2005-03-11T17:58:01.415647
      s LVL:  Info
      s HOST: yorda9-10.evl.uic.edu
      s METRIC: cpu_nice
      d VALUE: 0.0
      s PRG:  ganglia

...

Relevant Links

 

EVL ResearchGear publishes preliminary software, technical reports, data or results that the Electronic Visualization Laboratory openly shares with the research community. The work presented here is preliminary and we are not responsible for any damages that may result from its use or misuse. If you would like to cite any of this information in your research papers, presentations, etc, please reference the ResearchGear ID above. Thank you, and we hope you find the information on this page useful.


Visit ResearchGear Home Page