NAME

graffer — collect numeric values and generate graphs

SYNOPSIS

DESCRIPTION

The graffer utility can collect numeric values from external programs and produce graphs from the collected data.

The options are as follows:

-q

Query the current values from described external programs and store them in the database. This is usually done from a crontab(5) like this:

* * * * * /usr/local/bin/graffer -q

Note that without regular truncating (see below), the database will grow continually.

-p

Produce the configured set of graph images based on the statistics collected beforehand.

-g number:timeframe

Get stored values from the database for collect number according to the time frame and print them to stdout. Shows queue with last 16 or less calculated records, time frame maximum value, time frame average value, current maximum value and current average value according to the queue. Examples:

graffer -c /usr/local/etc/graffer/graffer.conf -g '7:from 3 hours to now'
graffer -c /usr/local/etc/graffer/graffer.conf -g '12:from 50 minutes to 30 minutes'
graffer -c /usr/local/etc/graffer/graffer.conf -g '23:from 52 weeks to 40 weeks'

-t days:[days]

Truncate the database, removing entries older than the specified number of days. The first value applies to uncompressed entries which constitute roughly 90 percent of the database, the second value to compressed entries. Uncompressed entries are needed only for high-resolution graphs over short time periods.

Assuming statistics are queried every I seconds, and a graph of width W pixels covering a time period of T seconds is generated, then uncompressed entries are only used when T/I < W*10.

In other words, removing uncompressed entries earlier than compressed ones drastically reduces the database size. Graphs can be generated over the entire time range the compressed entries cover (monthly or yearly graphs), and high resolution graphs over short time periods in the recent past don't lose any precision. The precision is gradually lost for older periods.

A typical invokation would be

graffer -t 31:365

This removes uncompressed entries older than a month and compressed entries older than a year, allowing generation of graphs up to one year into the past, while keeping full precision for one month.

Note that truncating the database does not usually shrink the size of the database file. Removed entries are only marked as deleted and their space is later reused for new entries.

-f file

Copy the database into a new file, skipping removed and invalid entries. This operation takes longer than truncating, but produces a smaller database file. This can also be used to recover data from a corrupted database, indicated by the following error during truncating:

data_truncate: db->del() returned 1
main: data_truncate() failed

-c config

Use the specified configuration file instead of the default /usr/local/etc/graffer/graffer.conf. Syntax:

collect = "collect" number = coldef .
coldef  = ( "path to external program" ) [ "tdiff" | "vdiff"].
image   = "image" filename "{"
              time theme size [ left ] [ right ] "}" .
time    = "from" number [ unit ] [ "to" number [ unit ] ] .
unit    = "minutes" | "hours" | "days" | "weeks" | "months" |
                      "years" .
theme	= "theme" ( "white" | "black" ) .
size    = "width" number "height" number .
left    = "left" graphs .
right   = "right" graphs .
graphs  = graph [ "," graphs ] .
graph   = "graph" number [ "bps" ] [ "avg" | "min" | "max" ]
                  label unit "color" red green blue [ "filled" ] .

collect lines define what values to store in the database. Each definition should use a unique number, which is used in graph lines to reference those values.

The tdiff option causes differential values (change of value per second) to be stored instead of absolute values. For example, storing interface byte counters (which count the number of bytes since last reset) differentially produces values representing current speed (bytes per second).

Several images can be defined in the same config file. An image can include two independent y-axes, both of which auto-scale independently to the maximum value in the selected range. Each axis can display an arbitrary number of graphs.

The vdiff option causes absolute differential values to be stored instead of absolute values. For example, storing interface packet counters (which count the number of packets since last reset).

When the bps option is used, values are multiplied by eight, and the unit prefixes 'k' (kilo), 'm' (mega), etc. are multiples of 1024, instead of 1000.

-C configdir

Config directory. Use all files from this directory instead of the default /usr/local/etc/graffer/graffer.conf.

-d database

Database file. Default is /var/db/graffer.db.

EXAMPLES

collect 1 = "/usr/local/bin/statgrab -u net.sis0.rx" tdiff
collect 2 = "/usr/local/bin/statgrab -u net.sis0.tx" tdiff
image "/var/www/htdocs/bandwidth.png" {
        from 7 days to now
        width 1000 height 400
        left
                graph 1 bps "in" "bits/s" color 0 192 0 filled
        right
                graph 2 bps "out" "bits/s" color 0 0 255
}
collect 3 = "/usr/local/etc/graffer/scripts/freebsd/netstat.sh rl0 in" tdiff
collect 4 = "/usr/local/etc/graffer/scripts/freebsd/disk.sh /dev/ad0 temp"
image "/var/www/htdocs/rl0.png" {
	from 12 months to now
	theme black
	width 800 height 200
	left
		graph 3 bps "states" "entries" color 200 0 0 filled,
	right
		graph 4 "ad0" "temperature" color 0 0 255
}

SEE ALSO

crontab(5)