If you want to display averages, maxima, percentiles, etcetera it is best to collect them now using the variable definition statement. Currently this makes no difference, but in a future version of rrdtool you may want to collect these values before consolidation.
The data fetched from the RRA is then consolidated so that there is exactly one datapoint per pixel in the graph. If you do not take care yourself, RRDtool will expand the range slightly if necessary. Note, in that case the first and/or last pixel may very well become unknown!
Sometimes data is not exactly in the format you would like to display it. For instance, you might be collecting bytes per second, but want to display bits per second. This is what the data calculation command is designed for. After consolidating the data, a copy is made and this copy is modified using a rather powerful RPN command set.
When you are done fetching and processing the data, it is time to graph it (or print it). This ends the rrdtool graph sequence.
filename can be '"-"' to send the image to "stdout". In this case, no other output is generated.
The start and end of the time series you would like to display, and which RRA the data should come from. Defaults are: 1 day ago until now, with the best possible resolution. Start and end can be specified in several formats, see AT-STYLE TIME SPECIFICATION and rrdgraph_examples. By default, rrdtool graph calculates the width of one pixel in the time domain and tries to get data from an RRA with that resolution. With the step option you can alter this behaviour. If you want rrdtool graph to get data at a one-hour resolution from the RRD, set step to 3'600. Note: a step smaller than one pixel will silently be ignored.
A horizontal string at the top of the graph and/or a vertically placed string at the left hand side of the graph.
A second axis will be drawn to the right of the graph. It is tied to the left axis via the scale and shift parameters. You can also define a label for the right axis.
By default the format of the axis lables gets determined automatically. If you want todo this your self, use this option with the same %lf arguments you know from the PRING and GPRINT commands.
By default, the width and height of the canvas (the part with the actual data and such). This defaults to 400 pixels by 100 pixels.
If you specify the --full-size-mode option, the width and height specify the final dimensions of the output image and the canvas is automatically resized to fit.
If you specify the --only-graph option and set the height < 32 pixels you will get a tiny graph image (thumbnail) to use as an icon for use in an overview, for example. All labeling will be stripped off the graph.
By default the graph will be autoscaling so that it will adjust the y-axis to the range of the data. You can change this behaviour by explicitly setting the limits. The displayed y-axis will then range at least from lower-limit to upper-limit. Autoscaling will still permit those boundaries to be stretched unless the rigid option is set.
Sometimes the default algorithm for selecting the y-axis scale is not satisfactory. Normally the scale is selected from a predefined set of ranges and this fails miserably when you need to graph something like "260 + 0.001 * sin(x)". This option calculates the minimum and maximum y-axis from the actual minimum and maximum data values. Our example would display slightly less than "260-0.001" to slightly more than "260+0.001" (this feature was contributed by Sasha Mikheev).
Where "--alt-autoscale" will modify both the absolute maximum AND minimum values, this option will only affect the minimum value. The maximum value, if not defined on the command line, will be 0. This option can be useful when graphing router traffic when the WAN line uses compression, and thus the throughput may be higher than the WAN line speed.
Where "--alt-autoscale" will modify both the absolute maximum AND minimum values, this option will only affect the maximum value. The minimum value, if not defined on the command line, will be 0. This option can be useful when graphing router traffic when the WAN line uses compression, and thus the throughput may be higher than the WAN line speed.
In order to avoid anti-aliasing blurring effects rrdtool snaps points to device resolution pixels, this results in a crisper appearance. If this is not to your liking, you can use this switch to turn this behaviour off.
Gridfitting is turned off for PDF, EPS, SVG output by default.
The x-axis label is quite complex to configure. If you don't have very special needs it is probably best to rely on the autoconfiguration to get this right. You can specify the string "none" to suppress the grid and labels altogether.
The grid is defined by specifying a certain amount of time in the ?TM positions. You can choose from "SECOND", "MINUTE", "HOUR", "DAY", "WEEK", "MONTH" or "YEAR". Then you define how many of these should pass between each line or label. This pair (?TM:?ST) needs to be specified for the base grid (G??), the major grid (M??) and the labels (L??). For the labels you also must define a precision in LPR and a strftime format string in LFM. LPR defines where each label will be placed. If it is zero, the label will be placed right under the corresponding line (useful for hours, dates etcetera). If you specify a number of seconds here the label is centered on this interval (useful for Monday, January etcetera).
This places grid lines every 10 minutes, major grid lines every hour, and labels every 4 hours. The labels are placed under the major grid lines as they specify exactly that time.
This places grid lines every 8 hours, major grid lines and labels each day. The labels are placed exactly between two major grid lines as they specify the complete day and not just midnight.
Y-axis grid lines appear at each grid step interval. Labels are placed every label factor lines. You can specify "-y none" to suppress the grid and labels altogether. The default for this option is to automatically select sensible values.
If you have set --y-grid to 'none' not only the labels get suppressed, also the space reserved for the labels is removed. You can still add space manually if you use the --units-length command to explicitly reserve space.
Place the Y grid dynamically based on the graph's Y range. The algorithm ensures that you always have a grid, that there are enough but not too many grid lines, and that the grid is metric. That is the grid lines are placed every 1, 2, 5 or 10 units. This parameter will also ensure that you get enough decimals displayed even if your graph goes from 69.998 to 70.001. (contributed by Sasha Mikheev).
Logarithmic y-axis scaling.
This sets the 10**exponent scaling of the y-axis values. Normally, values will be scaled to the appropriate units (k, M, etc.). However, you may wish to display units always in k (Kilo, 10e3) even if the data is in the M (Mega, 10e6) range, for instance. Value should be an integer which is a multiple of 3 between -18 and 18 inclusively. It is the exponent on the units you wish to use. For example, use 3 to display the y-axis values in k (Kilo, 10e3, thousands), use -6 to display the y-axis values in u (Micro, 10e-6, millionths). Use a value of 0 to prevent any scaling of the y-axis values.
This option is very effective at confusing the heck out of the default rrdtool autoscaler and grid painter. If rrdtool detects that it is not successful in labeling the graph under the given circumstances, it will switch to the more robust --alt-y-grid mode.
How many digits should rrdtool assume the y-axis labels to be? You may have to use this option to make enough space once you start fideling with the y-axis labeling.
With this option y-axis values on logarithmic graphs will be scaled to the appropriate units (k, M, etc.) instead of using exponential notation. Note that for linear graphs, SI notation is used by default.
Only generate the graph if the current graph is out of date or not existent. Note, that all the calculations will happen regardless so that the output of PRINT and graphv will be complete regardless. Note that the behaviour of lazy in this regard has seen several changes over time. The only thing you can realy rely on before rrdtool 1.3.7 is that lazy will not generate the graph when it is already there and up to date, and also that it will output the size of the graph.
After the image has been created, the graph function uses printf together with this format string to create output similar to the PRINT function, only that the printf function is supplied with the parameters filename, xsize and ysize. In order to generate an IMG tag suitable for including the graph into a web page, the command line would look like this:
--imginfo '<IMG SRC="/img/%s" WIDTH="%lu" HEIGHT="%lu" ALT="Demo">'
Override the default colors for the standard elements of the graph. The COLORTAG is one of "BACK" background, "CANVAS" for the background of the actual graph, "SHADEA" for the left and top border, "SHADEB" for the right and bottom border, "GRID", "MGRID" for the major grid, "FONT" for the color of the font, "AXIS" for the axis of the graph, "FRAME" for the line around the color spots, and finally "ARROW" for the arrow head pointing up and forward. Each color is composed out of three hexadecimal numbers specifying its rgb color component (00 is off, FF is maximum) of red, green and blue. Optionally you may add another hexadecimal number specifying the transparency (FF is solid). You may set this option several times to alter multiple defaults.
A green arrow is made by: "--color ARROW#00FF00"
Zoom the graphics by the given amount. The factor must be > 0
This lets you customize which font to use for the various text elements on the RRD graphs. "DEFAULT" sets the default value for all elements, "TITLE" for the title, "AXIS" for the axis labels, "UNIT" for the vertical unit label, "LEGEND" for the graph legend, "WATERMARK" for the watermark on the edge of the graph.
Use Times for the title: "--font TITLE:13:Times"
If you do not give a font string you can modify just the size of the default font: "--font TITLE:13:".
If you specify the size 0 then you can modify just the font without touching the size. This is especially useful for altering the default font without resetting the default fontsizes: "--font DEFAULT:0:Courier".
RRDtool comes with a preset default font. You can set the environment variable "RRD_DEFAULT_FONT" if you want to change this.
RRDtool uses Pango for its font handling. This means you can to use the full Pango syntax when selecting your font:
The font name has the form "[FAMILY-LIST] [STYLE-OPTIONS] [SIZE]", where FAMILY-LIST is a comma separated list of families optionally terminated by a comma, STYLE_OPTIONS is a whitespace separated list of words where each WORD describes one of style, variant, weight, stretch, or gravity, and SIZE is a decimal number (size in points) or optionally followed by the unit modifier ``px'' for absolute size. Any one of the options may be absent.
There are 3 font render modes:
normal: Full Hinting and Antialiasing (default)
light: Slight Hinting and Antialiasing
mono: Full Hinting and NO Antialiasing
(this gets ignored in 1.3 for now!)
This specifies the largest font size which will be rendered bitmapped, that is, without any font smoothing. By default, no text is rendered bitmapped.
All text in rrdtool is rendered using Pango. With the --pango-markup option, all text will be processed by pango markup. This allows to embed some simple html like markup tags using
Apart from the verbose syntax, there are also the following short tags available.
b Bold big Makes font relatively larger, equivalent to <span size="larger"> i Italic s Strikethrough sub Subscript sup Superscript small Makes font relatively smaller, equivalent to <span size="smaller"> tt Monospace font u Underline
More details on <http://developer.gnome.org/doc/API/2.0/pango/PangoMarkupFormat.html>.
There are 2 render modes:
normal: Graphs are fully Antialiased (default)
mono: No Antialiasing
RRDtool graphs are composed of stair case curves by default. This is in line with the way RRDtool calculates its data. Some people favor a more 'organic' look for their graphs even though it is not all that true.
Image format for the generated graph. For the vector formats you can choose among the standard Postscript fonts Courier-Bold, Courier-BoldOblique, Courier-Oblique, Courier, Helvetica-Bold, Helvetica-BoldOblique, Helvetica-Oblique, Helvetica, Symbol, Times-Bold, Times-BoldItalic, Times-Italic, Times-Roman, and ZapfDingbats.
(this gets ignored in 1.3 for now!)
If images are interlaced they become visible on browsers more quickly.
Suppress generation of the legend; only render the graph.
Force the generation of HRULE and VRULE legends even if those HRULE or VRULE will not be drawn because out of graph boundaries (mimics behaviour of pre 1.0.42 versions).
By default the tab-width is 40 pixels, use this option to change it.
If you are graphing memory (and NOT network traffic) this switch should be set to 1024 so that one Kb is 1024 byte. For traffic measurement, 1 kb/s is 1000 b/s.
Adds the given string as a watermark, horizontally centered, at the bottom of the graph.
You need at least one DEF statement to generate anything. The other statements are useful but optional. See rrdgraph_data and rrdgraph_rpn for the exact format.
NOTE: Graph and print elements
You need at least one graph element to generate an image and/or at least one print statement to generate a report. See rrdgraph_graph for the exact format.
When the filename '-' is given, the contents of the graph itself will also be returned through this interface (hash key 'image'). On the command line the output will look like this:
print = "0.020833" print = "0.0440833" graph_left = 51 graph_top = 22 graph_width = 400 graph_height = 100 graph_start = 1232908800 graph_end = 1232914200 image_width = 481 image_height = 154 value_min = 0.0000000000e+00 value_max = 4.0000000000e-02 image = BLOB_SIZE:8196 [... 8196 bytes of image data ...]
There is more information returned than in the standard interface. Especially the 'graph_*' keys are new. They help applications that want to know what is where on the graph.
Make sure to read rrdgraph_examples for tips&tricks.
This manual page by Alex van den Bogaerdt <email@example.com> with corrections and/or additions by several people