com.bigdata.counters.query
Class CounterSetQuery

java.lang.Object
  com.bigdata.counters.query.CounterSetQuery

public class CounterSetQuery
extends Object

Utility to extract a batch of performance counters from a collection of logged XML counter set files. This utility accepts file(s) giving the URLs which would be used to demand the corresponding performance counters against the live bigdata federation. The URLs listed in that file are parsed. The host and port information are ignored, but the URL query parameters are extracted and used to configured a set of URLQueryModels.

A single pass is made through the specified XML counter set files. Each file is read into memory by itself, and each query implied by a listed URL is run against the in-memory CounterSet hierarchy. The results are collected in independent CounterSet provisioned for the specified reporting units, etc. Once the last XML counter set file has been processed, the various reports requested by the listed URLs are generated.

For each generated report, the name of the file on which the report will be written is taken from the name of the counter whose value was extracted for that report. This filename may be overridden by including the URL query parameter , which specifies the file on which to write the report for that query.

Version:

$Id: CounterSetQuery.java 4702 2011-06-13 16:25:38Z thompsonbry $

Author:

Bryan Thompson

See Also:

IHostCounters, IRequiredHostCounters

TODO:

When rendering HTML output using flot, the flot resources need to be available in order to view the graphs. They should be written once into the output directory and the links in the (X)HTML output should resolve them there., Permit nsamples to be specified in units of minutes, hours, days. E.g., 3d would be 3 days., Performance for long runs could be improved if we use more efficient classes for mutable strings in XMLUtility and perhaps CounterSet. E.g., mg4j mutable string or http://javolution.org/.

Constructor Summary

CounterSetQuery()

Method Summary

static void

main(String[] args) Utility class for running extracting data from performance counter dumps and running various kinds of reports on those data.

Methods inherited from class java.lang.Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Constructor Detail

CounterSetQuery

public CounterSetQuery()

Method Detail

main

public static void main(String[] args)
                 throws IOException,
                        SAXException,
                        ParserConfigurationException,
                        InterruptedException,
                        ExecutionException

Utility class for running extracting data from performance counter dumps and running various kinds of reports on those data.

Usage:

-outputDir

The output directory (default is the current working directory).

-mimeType

The default MIME type for the rendered reports. The default is text/plain, but can be overridden on a query by query basis using URLQueryModel.MIMETYPE.

-nsamples

Override for the default #of history samples to be retained. It is an error if there are more distinct samples in the processed XML counter set files (that is, if the #of time periods sampled exceeds this many samples). If there are fewer, then some internal arrays will be dimensioned larger than is otherwise necessary.

-events <file>

A file containing tab-delimited Events. The Events are not required for simple performance counter views.

-queries <file>

A file, or directory of files, containing a list of URLs, each of which is interpreted as a URLQueryModel.

<file>(s)

One or more XML counter set files or directories containing such files. All such files will be processed before the reports are generated.

Parameters:

args - Command line arguments.

Throws:

IOException

ParserConfigurationException

SAXException

ExecutionException

InterruptedException