Introduction

QCG Application Monitoring service uses monitoring scripts to get information about current application computation status. These information is then rendered in accordance with the render scheme. Render scheme defines the graphical layout and transformation of the received data to the form of text, graphs and tables. For the convenience of users, several schemes were created, and we are open to create new schemas. For each render scheme there is monitoring script which parses application's output and temporary files.

Monitoring script

Monitoring script is an appliaction binary or script (bash, perl, awk, pyhon etc) that ca be run in linux environment. The script is executed periodically at certain period of time in a working directory of monitored application. The purpose of monitoring script is to write to standard output information about the current state of calculations. These information can be retrieved from the output or temporary files or application environment. The data format written out by the monitoring script is strictly dependent on the render scheme. The first line of output is common to all render schemas, and must have a form:

qcg.monitor.scheme={scheme_name}

, when {scheme_name} is name of the render scheme.

Schemas

In following sections we presental currently available schemas.

generic

This monitoring scheme displays all lines received from monitoring script along with time of receiving notification.

Sample monitoring script

generic2

The generic2 monitoring scheme display all lines received from last monitoring script notification. All previous data sent by monitoring script is ignored.

Sample monitoring script

gaussian

The monitoring scheme for Gaussian application, presents following information:

• energy from all optimization steps,
• energy difference between following optimization steps,
• graph with energy in following steps,
• graph with energy from last steps.

Monitoring script .

gaussianext

The extended monitoring scheme for Gaussian application, presents following information:

• number of optimization steps,
• current energy,
• last maximum force,
• last RMS force,
• last max displacement,
• last RMS displacement,
• graph with energy value in following steps,
• graph with energy from last steps,
• graph with maximum and RMS force in following steps,
• graph with maximum and RMS displacement in following steps.
• table with energy and energy difference values in following steps.

Monitoring script .

piernik

piernik monitoring scheme was created for a custom user application. In each step, application computes values for over a dozen parameters. Monitoring script sends a table with these parameters, and render scheme enables the user to select the parameters for which graph should be generated.

Monitoring script .

Job submission

QCG Application Monitoring service is integrated with the other QCG services. To submit a job with monitoring using QCG Simple Client, the watch-output directive must be used in the job description. In addition, the monitoring script must be stage in.

The sample job description for the gaussian monitoring scheme:

#!/bin/bash
 
#QCG stage-in-file=Naphthalene.gjf
#QCG stage-in-file=gaussian.ntf
 
#QCG watch-output=20,gaussian.ntf
 
#QCG application=g09
#QCG argument=Naphthalene.gjf

Detail description of QCG Simple Client can be found at  link.