The Perl Toolchain Summit needs more sponsors. If your company depends on Perl, please support this very important event.

NAME

RRDTool::OO - Object-oriented interface to RRDTool

SYNOPSIS

    use RRDTool::OO;

        # Constructor     
    my $rrd = RRDTool::OO->new(
                 file => "myrrdfile.rdd" );

        # Create a round-robin database
    $rrd->create(
         step        => 1,  # one-second intervals
         data_source => { name      => "mydatasource",
                          type      => "GAUGE" },
         archive     => { rows      => 5 });

        # Update RRD with sample values, use current time.
    for(1..5) {
        $rrd->update($_);
        sleep(1);
    }

        # Start fetching values from one day back, 
        # but skip undefined ones first
    $rrd->fetch_start();
    $rrd->fetch_skip_undef();

        # Fetch stored values
    while(my($time, $value) = $rrd->fetch_next()) {
         print "$time: ", 
               defined $value ? $value : "[undef]", "\n";
    }

        # Draw a graph in a PNG image
    $rrd->graph(
      image          => "mygraph.png",
      vertical_label => 'My Salary',
      start          => time() - 10,
      draw           => {
          type   => "area",
          color  => '0000FF',
          legend => "Salary over Time",
      }
    );

DESCRIPTION

RRDTool::OO is an object-oriented interface to Tobi Oetiker's round robin database tool rrdtool. It uses rrdtool's RRDs module to get access to rrdtool's shared library.

RRDTool::OO tries to marry rrdtool's database engine with the dwimminess and whipuptitude Perl programmers take for granted. Using RRDTool::OO abstracts away implementation details of the RRD engine, uses easy to memorize named parameters and sets meaningful defaults for parameters not needed in simple cases. For the experienced user, however, it provides full access to rrdtool's API. (Please check "Development Status" to verify how much of it has been implemented yet, though, since this module is under development :).

FUNCTIONS

my $rrd = RRDTool::OO->new( file => $file )

The constructor hooks up with an existing RRD database file $file, but doesn't create a new one if none exists. That's what the create() methode is for. Returns a RRDTool::OO object, which can be used to get access to the following methods.

$rrd->create( ... )

Creates a new round robin database (RRD). A RRD consists of one or more data sources and one or more archives:

    $rrd->create(
         step        => 60,
         data_source => { name      => "mydatasource",
                          type      => "GAUGE" },
         archive     => { rows      => 5 });

This defines a RRD database with a step rate of 60 seconds in between primary data points. Additionally, the RRD start time can be specified by specifying a start parameter.

It also sets up one data source named my_data_source of type GAUGE, telling rrdtool to use values of data samples as-is, without additional trickery.

And it creates a single archive with a 1:1 mapping between primary data points and archive points, with a capacity to hold five data points.

The RRD's step parameter is optional, and will be set to 300 seconds by rrdtool by default.

In addition to the mandatory settings for name and type, data_source parameter takes the following optional parameters: min (minimum input, defaults to U), max (maximum input, defaults to U), heartbeat (defaults to twice the RRD's step rate).

Archives expect at least one parameter, rows indicating the number of data points the archive is configured to hold. If nothing else is set, rrdtool will store primary data points 1:1 in the archive.

If you want to combine several primary data points into one archive point, specify values for cpoints (the number of points to combine) and cfunc (the consolidation function) explicitely:

    $rrd->create(
         step        => 60,
         data_source => { name      => "mydatasource",
                          type      => "GAUGE" },
         archive     => { rows      => 5,
                          cpoints   => 10,
                          cfunc     => 'AVERAGE',
                        });

This will collect 10 data points to form one archive point, using the calculated average, as indicated by the parameter cfunc (Consolidation Function, CF). Other options for cfunc are MIN, MAX, and LAST.

If you're defining multiple data sources or multiple archives, just provide them in this manner:

       # Define the RRD
    my $rc = $rrd->create(
        step        => 60,
        data_source => { name      => 'load1',
                         type      => 'GAUGE',
                       },
        data_source => { name      => 'load2',
                         type      => 'GAUGE',
                       },
        archive     => { rows      => 5,
                         cpoints   => 10,
                         cfunc     => 'AVERAGE',
                        },
        archive     => { rows      => 5,
                         cpoints   => 10,
                         cfunc     => 'MAX',
                        },
    );
$rrd->update( ... )

Update the round robin database with a new data sample, consisting of a value and an optional time stamp. If called with a single parameter, like in

    $rrd->update($value);

then the current timestamp and the defined $value will be used. If update is called with a named parameter list like in

    $rrd->update(time => $time, value => $value);

then the given timestamp $time is used along with the given value $value.

When updating multiple data sources, use the values parameter (instead of value) and pass an arrayref:

    $rrd->update(time => $time, values => [$val1, $val2, ...]);

This way, rrdtool expects you to pass in the data values in exactly the same order as the data sources were defined in the create method. If that's not the case, then the values parameter also accepts a hashref, mapping data source names to values:

    $rrd->update(time => $time, 
                 values => { $dsname1 => $val1, 
                             $dsname2 => $val2, ...});

RRDTool::OO will transform this automagically into RRDTool's template syntax.

$rrd->fetch_start( ... )

Initializes the iterator to fetch data from the RRD. This works nicely without any parameters if your archives are using a single consolidation function (e.g. MAX). If there's several archives in the RRD using different consolidation functions, you have to specify which one you want:

    $rrd->fetch_start(cfunc => "MAX");

Other options for cfunc are MIN, AVERAGE, and LAST.

fetch_start features a number of optional parameters: start, end and resolution.

If the start time parameter is omitted, the fetch starts 24 hours before the end of the archive. Also, an end time can be specified:

    $rrd->fetch_start(start => time()-10*60,
                      end   => time());

The third optional parameter, resolution defaults to the highest resolution available and can be set to a value in seconds, specifying the time interval between the data samples extracted from the RRD. See the rrdtool fetch manual page for details.

Development note: The current implementation fetches all values from the RRA in one swoop and caches them in memory. This might change in the future, to cache only the last timestamp and keep fetching from the RRD with every fetch_next() call.

$rrd->fetch_skip_undef()

rrdtool doesn't remember the time the first data sample went into the archive. So if you run a rrdtool fetch with a start time of 24 hours ago and you've only submitted a couple of samples to the archive, you'll see many undef values.

Starting from the current iterator position (or at the specified start time immediately after a fetch_start()), fetch_skip_undef() will skip all undef values in the RRA and positions the iterator right before the first defined value. If all values in the RRA are undefined, the a following $rrd->fetch_next() will return undef.

($time, $value, ...) = $rrd->fetch_next()

Gets the next row from the RRD iterator, initialized by a previous call to $rrd->fetch_start(). Returns the time of the archive point along with all values as a list.

$rrd->graph( ... )

If there's only one data source in the RRD, drawing a nice graph in an image file on disk is as easy as

    $rrd->graph(
      image          => $image_file_name,
      vertical_label => 'My Salary',
      draw           => { thickness => 2,
                          color     => 'FF0000',
                          legend    => 'Salary over Time',
                        },
    );

This will assume a start time of 24 hours before now and an end time of now. Specify start and end explicitely to be clear:

    $rrd->graph(
      image          => $image_file_name,
      vertical_label => 'My Salary',
      start          => time() - 24*3600,
      end            => time(),
      draw           => { thickness => 2,
                          color     => 'FF0000',
                          legend    => 'Salary over Time',
                        },
    );

As always, RRDTool::OO will pick reasonable defaults for parameters not specified. The values for data source and consolidation function default to the first values it finds in the RRD. If there are multiple datasources in the RRD or multiple archives with different values for cfunc, just specify explicitely which one to draw:

    $rrd->graph(
      image          => $image_file_name,
      vertical_label => 'My Salary',
      draw           => {
        thickness => 2,
        color     => 'FF0000',
        dsname    => "load",
        cfunc     => 'MAX'},
    );

If draw doesn't define a type, it defaults to "line". If you don't want to define a type (because the graph shouldn't drawn), use type => "hidden". Other values are "area" for solid colored areas and "stack" for graphical values stacked on top of each other. And you can certainly have more than one graph in the picture:

    $rrd->graph(
      image          => $image_file_name,
      vertical_label => 'My Salary',
      draw           => {
        type      => 'area',
        color     => 'FF0000', # red area
        dsname    => "load",
        cfunc     => 'MAX'},
      draw        => {
        type      => 'stack',
        color     => '00FF00', # a green area stacked on top of the red one 
        dsname    => "load",
        cfunc     => 'AVERAGE'},
    );

Graphs may assemble data from different RRD files. Just specify which file you want to draw the data from, using draw:

    $rrd->graph(
      image          => $image_file_name,
      vertical_label => 'Network Traffic',
      draw           => {
        file      => "file1.rrd",
        legend    => "First Source",
      },
      draw        => {
        file      => "file2.rrd",
        type      => 'stack',
        color     => '00FF00', # a green area stacked on top of the red one 
        dsname    => "load",
        legend    => "Second Source",
        cfunc     => 'AVERAGE'
      },
    );

If a file parameter is specified per draw, the defaults for dsname and cfunc are fetched from this file, not from the file that's attached to the RRDTool::OO object $rrd used.

Graphs may also consist of algebraic calculations of previously defined graphs. In this case, graphs derived from real data sources need to be named, so that subsequent cdef definitions can refer to them and calculate new graphs, based on the previously defined graph:

    $rrd->graph(
      image          => $image_file_name,
      vertical_label => 'Network Traffic',
      draw           => {
        type      => 'line',
        color     => 'FF0000', # red line
        dsname    => 'load',
        name      => 'firstgraph',
        legend    => 'Unmodified Load',
      },
      draw        => {
        type      => 'line',
        color     => '00FF00', # green line
        cdef      => "firstgraph,2,*",
        legend    => 'Load Doubled Up',
      },
    );

Note that the second draw doesn't refer to a datasource dsname (nor does it fall back to the default data source), but defines a cdef, performing calculations on a previously defined draw named firstgraph. The calculation is specified using RRDTool's reverse polish notation, where instructions are separated by commas ("firstgraph,2,*" simply multiplies firstgraph's values by 2).

On a global level, in addition to the vertical_label parameter shown in the examples above, graph offers a plethora of parameters:

vertical_label, title, start, end, x_grid, y_grid, alt_y_grid, no_minor, alt_y_mrtg, alt_autoscale, alt_autoscale_max, units_exponent, units_length, width, height, interlaced, imginfo, imgformat, overlay, unit, lazy, upper_limit, logarithmic, color, no_legend, only_graph, force_rules_legend, title, step.

Some options (e.g. alt_y_grid) don't expect values, they need to be specified like

    alt_y_grid => undef

in order to be passed properly to RRDTool.

The color option expects a reference to a hash with various settings for the different graph areas: back (background), canvas, shadea (left/top border), shadeb (right/bottom border), grid, mgrid major grid, font, frame and arrow:

    $rrd->graph(
      ...
      color          => { back   => '#0e0e0e',
                          arrow  => '#ff0000',
                          canvas => '#eebbbb',
                        },
      ...
    );

Please check the RRDTool documentation for a detailed description on what each option is used for:

    http://people.ee.ethz.ch/~oetiker/webtools/rrdtool/manual/rrdgraph.html

Sometimes it's useful to print max, min or average values of a given graph at the bottom of the chart or to STDOUT. That's what gprint and print options are for. In addition to the draw name (defaults to the first draw), a consolidation function can be specified: MIN, MAX, AVERAGE, LAST (defaults to AVERAGE). Note that this is unrelated to the data source's consolidation function, it's just applied on the graph print/gprint refer to. Finally, the format parameter gives a printf-like template (defaults to "Average=%lf": A call to

    $rrd->graph(
      image          => $image_file_name,
      draw           => {
        name      => "first_draw",
        dsname    => "load",
        cfunc     => 'MAX'},
      gprint         => {
        draw      => 'first_draw',
        cfunc     => 'AVERAGE',
        format    => 'Average=%lf',
      },
    );

prints "Average=x.xx" at the bottom of the graph, showing what the average value of the graph is.

$rrd->dump()

Available as of rrdtool 1.0.49.

Dumps the RRD in XML format to STDOUT. If you want to dump it into a file instead, do this:

    my $pid;

    unless ($pid = open DUMP, "-|") {
      die "Can't fork: $!" unless defined $pid;
      $rrd->dump();
      exit 0;
    }

    waitpid($pid, 0);

    open OUT, ">out";
    print OUT $_ for <DUMP>;
    close OUT;
my $hashref = $rrd->info()

Grabs the RRD's meta data and returns it as a hashref, holding a map of parameter names and their values.

my $time = $rrd->last()

Return the RRD's last update time.

$rrd->restore(xml => "file.xml")

Available as of rrdtool 1.0.49.

Restore a RRD from a dump. The xml parameter specifies the name of the XML file containing the dump. If the optional flag range_check is set to a true value, restore will make sure the values in the RRAs do not exceed the limits defined for the different datasources:

    $rrd->restore(xml => "file.xml", range_check => 1);
$rrd->tune( ... )

Alter a RRD's data source configuration values:

        # Set the heartbeat of the RRD's only datasource to 100
    $rrd->tune(heartbeat => 100);

        # Set the minimum of DS 'load' to 1
    $rrd->tune(dsname => 'load', minimum => 1);

        # Set the maximum of DS 'load' to 10
    $rrd->tune(dsname => 'load', maximum => 10);

        # Set the type of DS 'load' to AVERAGE
    $rrd->tune(dsname => 'load', type => 'AVERAGE');

        # Set the name of DS 'load' to 'load2'
    $rrd->tune(dsname => 'load', name => 'load2');
$rrd->error_message()

Return the message of the last error that occurred while interacting with RRDTool::OO.

Development Status

The following methods are not yet implemented:

rrdresize, xport, rrdcgi.

Error Handling

By default, RRDTool::OO's methods will throw fatal errors (as in: they're calling die) if the underlying RRDs::* commands indicate failure.

This behaviour can be overridden by calling the constructor with the raise_error flag set to false:

    my $rrd = RRDTool::OO->new(
        file        => "myrrdfile.rdd",
        raise_error => 0,
    );

In this mode, RRDTool's methods will just pass back values returned from the underlying RRDs functions if an error happens (usually 1 if successful and undef if an error occurs).

Debugging

RRDTool::OO is Log::Log4perl enabled, so if you want to know what's going on under the hood, just turn it on:

    use Log::Log4perl qw(:easy);

    Log::Log4perl->easy_init({
        level    => $DEBUG
    }); 

If you're interested particularily in rrdtool commands issued by RRDTool::OO while you're operating it, just enable the category "rrdtool":

    Log::Log4perl->easy_init({
        level    => $INFO, 
        category => 'rrdtool',
        layout   => '%m%n',
    }); 

This will display all rrdtool commands that RRDTool::OO submits to the shared library. Let's turn it on for the code snippet in the SYNOPSIS section of this manual page and watch the output:

    rrdtool create myrrdfile.rdd --step 1 \
            DS:mydatasource:GAUGE:2:U:U RRA:MAX:0.5:1:5
    rrdtool update myrrdfile.rdd N:1
    rrdtool update myrrdfile.rdd N:2
    rrdtool update myrrdfile.rdd N:3
    rrdtool fetch myrrdfile.rdd MAX

Often handy for cut-and-paste.

INSTALLATION

RRDTool::OO requires a rrdtool installation with the RRDs Perl module, that comes with the rrdtool distribution.

Download the tarball from

    http://people.ee.ethz.ch/~oetiker/webtools/rrdtool/pub/rrdtool.tar.gz

and then unpack, compile and install:

    tar zxfv rrdtool.tar.gz
    cd rrdtool-1.0.46
    make
    cd perl-shared
    perl Makefile.PL
    ./configure
    make
    make test
    make install

SEE ALSO

  • Tobi Oetiker's RRDTool homepage at

        http://rrdtool.org

    especially the manual page at

            http://people.ee.ethz.ch/~oetiker/webtools/rrdtool/manual/index.html
  • My articles on rrdtool in "Linux Magazin" (Germany) and "Linux Magazine" (UK):

        http://www.linux-magazin.de/Artikel/ausgabe/2004/06/perl/perl.html
        http://www.linux-magazine.com/issue/44/Limiting_Data.pdf (not online yet)

AUTHOR

Mike Schilli, <m@perlmeister.com>

COPYRIGHT AND LICENSE

Copyright (C) 2004 by Mike Schilli

This library is free software; you can redistribute it and/or modify it under the same terms as Perl itself, either Perl version 5.8.3 or, at your option, any later version of Perl 5 you may have available.