NAME
rrdpoller - Retrieve RRD file data and apply some threshold algorithm
SYNOPSIS
rrdpoller [OPTIONS] list|ls *filename*
rrdpoller [OPTIONS] get *filename* *datasource* [offset=*time*]
[cf=*func-name*] [and|or command...]
rrdpoller [OPTIONS] exact *filename* *datasource* *value* [and|or
command...]
rrdpoller [OPTIONS] boundaries *filename* *datasource* [min=*number*]
[max=*number*] [and|or command...]
rrdpoller [OPTIONS] relation *filename* *datasource*
[*<*|*>*]*threshold*[*%*] [target=*filename*] [compare-ds=*datasource*]
[offset=*time*] [and|or command...]
rrdpoller [OPTIONS] quotient *filename* *datasource*
[*<*|*>*]*threshold*[*%*] [target=*filename*] [target-ds=*datasource*]
[offset=*time*] [and|or command...]
rrdpoller [OPTIONS] hunt *filename* *datasource* *roll*
[parent=*filename*] [parent-ds=*datasource*] [and|or command...]
DESCRIPTION
rrdpoller allows you to poll RRD files and get current value of a given
datasource. Additionally it implement some advanced checks able to use
past values to decide if the current one is out of threshold or not.
The main purpose of this tool is to do the interface between an existing
monitoring system like BigBrother, Nagios or Mon and a set of RRD graphs
generated by some specialized data collectors like rrdcollect, Cricket
and such. Thus you can build a very modular monitoring/trend
architecture without the need of double polling.
EXAMPLES
Those examples are fictive, they doesn't represent real numbers. It's
just to let you figure out what you can do with each algorithm.
rrdpoller ls host1-net.rrd
Returns the whole list of datasources contained in the file.
rrdpoller get host1-load.rrd 1min_avg
Get the current value of the *1min_avg* datasource of the
*host1-load.rrd* file.
rrdpoller --proxy http://xmlrpc.monserver.com:8080/ get host1-load.rrd
1min_avg
Same as precedent but querying a remove daemon (see rrdpollerd).
rrdpoller exact host1-service.rrd http 1
Get the current value of the *http* datasource of the
*host1-load.rrd* file and return false if not equal to 1.
rrdpoller boundaries host1-load.rrd 5min_avg max=15
Get the current value of the *5min_avg* datasource of the
*host1-load.rrd* file and return false if greater than *15*.
rrdpoller relation host1-disk.rrd usage 100000 offset=1min
Compare the current *usage* datasource value of the *host1-disk.rrd*
file with the same datasource 1 minute ago and return false if the
delta is greater than 100000.
rrdpoller quotient host1-mem.rrd usage 90% target-ds=total
Compare the current *usage* datasource of the *host1-mem.rrd* file
and divide it by the *total* datasource of the same file. The
command returns false if the obtained percentage is greater than
*90%*.
rrdpoller hunt pop2.rrd users 40 target=pop1.rrd
If the *users* datasource of the *pop2.rrd* file is non-zero and the
same datasource of the *pop1.rrd* hasn't yet reach *40*, the command
will return false.
rrdpoller quotient host1-mem.rrd high_mem,low_mem,+ 90%
target-ds=low_total,high_total,+
Same as former quotient example but using RPN datasource feature.
Here the high_mem and the low_mem datasources will be summed then
the low_total and the high_total too before to do the quotient
computation.
rrdpoller different mysql-slave.rrd insert,update,delete,+ 0 or equal
mysql-master.rrd insert,update,delete,+ 0
First test will fail if the sum of inserts, updates and deletes is
equal to 0. If failed, the second test will verify if the same sum
is also equal to 0 on the master and will finally fail if not. Thus
we can verify that the MySQL replication is running.
COMMAND SYNTAX
The rrdpoller is followed by one of its sub-command and related
arguments (See below). Sub-commands arguments are key=*value* pair and
can alternatively be written as --key *value* pair for backward
compatibility. Since version 1.5.0, several sub-commands can be chained
together with special keywords "and" and "or". Each sub-commands return
a result, one by line in the order they have been called. When several
sub-commands are chained, all sub-commands are executed, even if the
first have failed and the second was chained with the "and" keyword. You
can change this behavior with the --lazy option, but don't expect to see
the result of sub-commands that doesn't have been executed then.
COMMON OPTIONS
--proxy *URL*
If you use this option, the rrdpoller command will turn into an XML-RPC
client and will ask send all RRD queries to a remote rrdpoller daemon
(see rrdpollerd for more information).
--max-last *seconds*
If this option is specified with an integer, the test will return false
if the last updated row is older than X *seconds*. The exit code will be
2 if this test fails instead of 1 for normal tests.
--lazy
If several sub-commands are chained with the "and" or "or" keywords,
don't exec all sub-command if the expression doesn't need it.
--quiet
Don't print sub-command results. Implies --lazy
SUB-COMMANDS
All sub-commands needs at least the too first arguments *filename* and
*datasource* (except for list command). Except for the get and list
commands (which doesn't perform a test), all sub-command will change the
rrdpoller return code to false (1) if the test fails. All sub-commands
will print the given *datasource* value of the given *filename* on the
standard output.
The *datasource* can be an RPN expression (see Math::RPN) as well as all
other datasource options. Note that the Math::RPN module have to be
installed in order to use this feature.
list|ls *filename*
This command list all datasources of the given file.
get *filename* *datasource* [OPTIONS]
This sub-command the last inserted value in the given *datasource* of
the given RRD *filename*. You can get an older value by giving an offset
that will be expressed as: <last_inserted_time> - <offset>.
options:
offset=*time*
Makes the get command to retrieve an earlier value. It substract the
value of *time* to the timestamp of the last value inserted in the
*datasource*. The *time* can be expressed in second (ie *60*) or in
RRD time reference specification format (see "TIME REFERENCE
SPECIFICATION" in rrdfetch).
cf=*func-name*
Default consolidation function is AVERAGE. This option allow you to
change it to whichever you want.
exact *filename* *datasource* *exact*
This threshold allows you to monitor the *datasource* for an *exact*
match. This is useful in cases where an enumerated (or boolean) SNMP
object instruments a condition where a transition to a specific state
requires attention. For example, a datasource might return either
true(1) or false(2), depending on whether or not a power supply has
failed.
boundaries *filename* *datasource* [OPTIONS]
This threshold takes too optional values, a *min*imum and a *max*imum
value. If the *datasource* strays outside of this interval, the test
fail.
options:
min=*number*
If current value is lower than *number*, the test fail
max=*number*
If current value is greater than *number*, the test fail
relation *filename* *datasource* *[<|>]threshold[%]* [OPTIONS]
A relation threshold considers the difference between two *datasources*
(possibly from different RRD *files*), or alternatively, the difference
between two temporally distinct values for the same *datasource*. The
difference can be expressed as absolute value, or as a percentage of the
second *datasource* (comparison) value. This difference is compared to a
*threshold* argument with either the greater than (>) or lesser than (<)
operator. The criteria fails when the expression (<absolute or relative
difference> <either greater-than or less-than> <threshold>) evaluates to
false.
The *threshold* number, optionally preceded by the greater than (>) or
less than (<) symbol, and optionally followed by the symbol percent (%).
If omitted, greater than is used by default and the expression:
difference > threshold, is evaluated. "<10%", ">1000", "50%", and "500"
are all examples of valid thresholds.
options:
target=*filename*
The path of the comparison RRD *file*. This argument is optional and
if omitted the first RRD *file* is also taken as the comparison
target.
target-ds=*datasource*
The name of the comparison *datasource*. This *datasource* must
belong to the comparison RRD file. This argument is optional and if
omitted the first *datasource* name is also taken as the comparison
datasource name. If the value is a number, the value is considered
as a fix value and is taked for the comparison.
offset=*time*
The temporal *offset* to go back from the first value to fetch the
comparison datasource value. Note that a data source value must
exist in the RRD file for that exact offset. If This argument is
optional and if omitted, it is set to 0. The *time* value can be
expressed in second (ie *60*) or in RRD time reference specification
format (see "TIME REFERENCE SPECIFICATION" in rrdfetch).
quotient *filename* *datasource* *[<|>]threshold[%]* [OPTIONS]
Quotient thresholds are similar to relation thresholds, except that they
consider the quotient of two data sources, or alternatively, the same
data source at two different time points. For a quotient monitor
threshold, the value of the first data source is computed as a
percentage of the second data source value (such as 10 (first
datasource) is 50% of 20 (second datasource)). This percentage is then
compared to a threshold argument with either the greater than (>) or
less than (<) operator. The criteria fails when the expression
(<percentage> <either greater-than or less-than> <threshold>) evaluates
to false.
The *threshold* number, optionally preceded by the greater than (>) or
less than (<) symbol and followed by the symbol percent (%). If omitted,
greater than is used by default and the expression: difference >
threshold, is evaluated. "<10%" and "50%" are all examples of valid
thresholds.
options:
target=*filename*
The path of the comparison RRD *file*. This argument is optional and
if omitted the first RRD *file* is also taken as the comparison
target.
target-ds=*datasource*
The name of the comparison *datasource*. This *datasource* must
belong to the comparison RRD file. This argument is optional and if
omitted the first *datasource* name is also taken as the comparison
datasource name. If the value is a number, the value is considered
as a fix value and is taked for the comparison.
offset=*time*
The temporal *offset* to go back from the first value to fetch the
comparison datasource value. Note that a data source value must
exist in the RRD file for that exact offset. If This argument is
optional and if omitted, it is set to 0. The *time* value can be
expressed in second (ie *60*) or in RRD time reference specification
format (see "TIME REFERENCE SPECIFICATION" in rrdfetch).
hunt *roll* [OPTIONS]
The hunt threshold is designed for the situation where the data source
serves as an overflow for another data source; that is, if one data
source (the parent) is at or near capacity, then traffic will begin to
appear on this (the monitored) data source. One application of hunt
monitor thresholds is to identify premature rollover in a set of modem
banks configured to hunt from one to the next. Specifically, the
criteria of the hunt monitor threshold fails if the value of the
monitored data source is non-zero and the current value of the parent
data source falls below a specified capacity threshold.
The *roll* argument is the threshold of the parent data source.
Generally this should be slightly less than the maximum capacity of the
target.
options:
parent=*filename*
The path of the comparison RRD *file*. This argument is optional and
if omitted the first RRD *file* is also taken as the comparison
target.
parent-ds=*datasource*
The name of the comparison *datasource*. This *datasource* must
belong to the comparison RRD file. This argument is optional and if
omitted the first *datasource* name is also taken as the comparison
datasource name. If the value is a number, the value is considered
as a fix value and is taked for the comparison.
EXIT STATUS
0 Successful test
1 Performed test returned a false value
2 The max-last was reached
255 An exception was thrown
AUTHOR
Olivier Poitrey <rs@rhapsodyk.net>
LICENCE
rrdpoller retrieves RRD file data and apply some threshold algorithm.
Copyright (C) 2004 Olivier Poitrey
This program is free software; you can redistribute it and/or modify it
under the terms of the GNU General Public License as published by the
Free Software Foundation; either version 2 of the License, or (at your
option) any later version.
This program is distributed in the hope that it will be useful, but
WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General
Public License for more details.
You should have received a copy of the GNU General Public License along
with this program; if not, write to the Free Software Foundation, Inc.,
59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
SEE ALSO
RRD::Query, RRD::Threshold, rrdtool, RRDs, rrdpollerd