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: - . 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 ( ) 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 ( ) 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 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