Code

For 1.4.2 release
[nagiosplug.git] / doc / developer-guidelines.sgml
index e6a864d1f30650ff279c63979fc425bf40ff8dfa..c531fbb0218683f912444b58c48abe8771c56afc 100644 (file)
-<!DOCTYPE book PUBLIC "-//OASIS//DTD DocBook V4.1//EN">
+<!DOCTYPE book PUBLIC "-//OASIS//DTD DocBook V4.1//EN" >
 <book>
   <title>Nagios Plug-in Developer Guidelines</title>
 
   <bookinfo>
     <authorgroup>
       <author>
-       <firstname>Karl</firstname>
-       <surname>DeBisschop</surname>
-       <affiliation>
-         <address><email>karl@debisschop.net</email></address>
-       </affiliation>
+        <affiliation>
+          <orgname>Nagios Plugins Development Team</orgname>
+        </affiliation>
       </author>
-
-      <author>
-       <firstname>Ethan</firstname>
-       <surname>Galstad</surname>
-       <authorblurb>
-         <para>Author of Nagios</para>
-         <para><ulink url="http://www.nagios.org"></ulink></para>
-       </authorblurb>
-       <affiliation>
-         <address><email>netsaint@linuxbox.com</email></address>
-       </affiliation>
-      </author>
-
-      <author>
-       <firstname>Hugo</firstname>
-       <surname>Gayosso</surname>
-       <affiliation>
-         <address><email>hgayosso@gnu.org</email></address>
-       </affiliation>
-      </author>
-
-         
-       <author>
-       <firstname>Subhendu</firstname>
-       <surname>Ghosh</surname>
-       <affiliation>
-               <address><email>sghosh@sourceforge.net</email></address>
-       </affiliation>
-       </author>
-       
-       <author>
-       <firstname>Stanley</firstname>
-       <surname>Hopcroft</surname>
-       <affiliation>
-               <address><email>stanleyhopcroft@sourceforge.net</email></address>
-       </affiliation>
-       </author>       
-
     </authorgroup>
 
-    <pubdate>2002</pubdate>
+    <pubdate>2005</pubdate>
     <title>Nagios plug-in development guidelines</title>
        
     <revhistory>
        <revision>
-          <revnumber>0.4</revnumber>
-          <date>2 May 2002</date>
+          <revnumber>$Revision$</revnumber>
+          <date>$Date$</date>
        </revision>
     </revhistory>
 
        <copyright>
-               <year>2000 2001 2002</year> 
-               <holder>Karl DeBisschop, Ethan Galstad, 
-               Hugo Gayosso, Stanley Hopcroft, Subhendu Ghosh</holder>
+               <year>2000 - 2005</year> 
+               <holder>Nagios Plugins Development Team</holder>
        </copyright>
 
 </bookinfo>
 
 
-  <preface id=preface>
-    <title>About the guidelines</title>
-
+<preface id="preface"><title>Preface</title>
     <para>The purpose of this guidelines is to provide a reference for
     the plug-in developers and encourage the standarization of the
     different kind of plug-ins: C, shell, perl, python, etc.</para>
 
-
-    <section> <title>Copyright</title>
-
-        <para>Nagios Plug-in Development Guidelines Copyright (C) 2000 2001
-               2002
-        Karl DeBisschop, Ethan Galstad, Hugo Gayosso, Stanley Hopcroft, 
-               Subhendu Ghosh</para>
+        <para>Nagios Plug-in Development Guidelines Copyright (C) 2000-2005
+        (Nagios Plugins Team)</para>
 
         <para>Permission is granted to make and distribute verbatim
         copies of this manual provided the copyright notice and this
         permission notice are preserved on all copies.</para>
 
-               <para>The plugins themselves are copyrighted by their respective
-               authors.</para>
-
-    </section>
+       <para>The plugins themselves are copyrighted by their respective
+       authors.</para>
 </preface>
 
 <article>
+<section id="DevRequirements"><title>Development platform requirements</title>
+       <para>
+       Nagios plugins are developed to the GNU standard, so any OS which is supported by GNU
+       should run the plugins. While the requirements for compiling the Nagios plugins release 
+       is very small, to develop from CVS needs additional software to be installed. These are the 
+       minimum levels of software required:
+
+       <literallayout>
+       gnu make 3.79
+       automake 1.8
+       autoconf 2.58
+       gettext 0.11.5
+       </literallayout>
+
+       To compile from CVS, after you have checked out the code, run:
+       <literallayout>
+       tools/setup
+       ./configure
+       make
+       make install
+       </literallayout>
+       </para>
+</section>
+
 <section id="PlugOutput"><title>Plugin Output for Nagios</title>
        
                <para>You should always print something to STDOUT that tells if the 
-               service is working or why its failing. Try to keep the output short - 
+               service is working or why it is failing. Try to keep the output short - 
                probably less that 80 characters. Remember that you ideally would like 
                the entire output to appear in a pager message, which will get chopped
                off after a certain length.</para>
                when it notifies contacts about potential problems. If you print
                multiple lines, you're out of luck. Remember, keep it short and
                to the point.</para>
+
+               <para>Output should be in the format:</para>
+               <literallayout>
+               METRIC STATUS: Information text
+               </literallayout>
+               <para>However, note that this is not a requirement of the API, so you cannot depend on this
+               being an accurate reflection of the status of the service - the status should always 
+               be determined by the return code.</para>
                </section>
 
                <section><title>Verbose output</title>
                -v options for additional verbosity, up to a maximum of 3. The standard
                type of output should be:</para>
 
-               <table id="verbose_levels"><title>Verbose output levels</title>
+               <table id="verboselevels"><title>Verbose output levels</title>
                        <tgroup cols="2">
                                <thead>
                                        <row>
                                </thead>
                                <tbody>
                                        <row>
-                                               <entry align=center><para>0</para></entry>
+                                               <entry align="center"><para>0</para></entry>
                                                <entry><para>Single line, minimal output. Summary</para></entry>
                                        </row>
                                        <row>
-                                               <entry align=center><para>1</para></entry>
+                                               <entry align="center"><para>1</para></entry>
                                                <entry><para>Single line, additional information (eg list processes that fail)</para></entry>
                                        </row>
                                        <row>
-                                               <entry align=center><para>2</para></entry>
+                                               <entry align="center"><para>2</para></entry>
                                                <entry><para>Multi line, configuration debug output (eg ps command used)</para></entry>
                                        </row>
                                        <row>
-                                               <entry align=center><para>3</para></entry>
+                                               <entry align="center"><para>3</para></entry>
                                                <entry><para>Lots of detail for plugin problem diagnosis</para></entry>
                                        </row>
                                </tbody>
                occurred and error code is 138 or 255 or some such number.  These
                are usually caused by plugins using system commands and having not 
                enough checks to catch unexpected output.  Developers should include a
-               default catch-all for system command output that returns an UNKOWN
+               default catch-all for system command output that returns an UNKNOWN
                return code.</para>
                
                <table id="ReturnCodes"><title>Plugin Return Codes</title>
                                </thead>
                                <tbody>
                                        <row>
-                                               <entry align=center><para>0</para></entry>
-                                               <entry valign=middle><para>OK</para></entry>
+                                               <entry align="center"><para>0</para></entry>
+                                               <entry valign="middle"><para>OK</para></entry>
                                                <entry><para>The plugin was able to check the service and it 
                                                appeared to be functioning properly</para></entry>
                                        </row>
                                        <row>
-                                               <entry align=center><para>1</para></entry>
-                                               <entry valign=middle><para>Warning</para></entry>
+                                               <entry align="center"><para>1</para></entry>
+                                               <entry valign="middle"><para>Warning</para></entry>
                                                <entry><para>The plugin was able to check the service, but it 
                                                appeared to be above some "warning" threshold or did not appear 
                                                to be working properly</para></entry>
                                        </row>
                                        <row>
-                                               <entry align=center><para>2</para></entry>
-                                               <entry valign=middle><para>Critical</para></entry>
+                                               <entry align="center"><para>2</para></entry>
+                                               <entry valign="middle"><para>Critical</para></entry>
                                                <entry><para>The plugin detected that either the service was not 
                                                running or it was above some "critical" threshold</para></entry>
                                        </row>
                                        <row>
-                                               <entry align=center><para>3</para></entry>
-                                               <entry valign=middle><para>Unknown</para></entry>
+                                               <entry align="center"><para>3</para></entry>
+                                               <entry valign="middle"><para>Unknown</para></entry>
                                                <entry><para>Invalid command line arguments were supplied to the 
                                                plugin or the plugin was unable to check the status of the given 
                                                hosts/service</para></entry>
       
                </section>
 
+               <section id="thresholdformat"><title>Threshold range format</title>
+               <para>Thresholds ranges define the warning and critical levels for plugins to 
+               alert on. The theory is that the plugin will do some sort of check which returns
+               back a numerical value, or metric, which is then compared to the warning and 
+               critical thresholds.
+               This is the generalised format for threshold ranges:</para>
+
+               <literallayout>
+               [@]start:end
+               </literallayout>
+       
+               <para>Notes:</para>
+               <orderedlist>
+               <listitem><para>start &gt; end</para>
+                       </listitem>
+               <listitem><para>start and ":" is not required if start=0</para>
+                       </listitem>
+               <listitem><para>if range is of format "start:" and end is not specified, 
+                       assume end is infinity</para>
+                       </listitem>
+               <listitem><para>to specify negative infinity, use "~"</para>
+                       </listitem>
+               <listitem><para>alert is raised if metric is outside start and end range
+                       (inclusive of endpoints)</para>
+                       </listitem>
+               <listitem><para>if range starts with "@", then alert if inside this range
+                       (inclusive of endpoints)</para>
+                       </listitem>
+               </orderedlist>
+               
+               <para>Note: Not all plugins are coded to expect ranges in this format. It is
+               planned for a future release to
+               provide standard libraries to parse and compare metrics against ranges. There
+               will also be some work in providing multiple metrics.</para>
+               </section>
+
+               <section><title>Performance data</title>
+               <para>Performance data is defined by Nagios as "everything after the | of the plugin output" -
+               please refer to Nagios documentation for information on capturing this data to logfiles.
+               However, it is the responsibility of the plugin writer to ensure the 
+               performance data is in a "Nagios plugins" format.
+               This is the expected format:</para>
+
+               <literallayout>
+               'label'=value[UOM];[warn];[crit];[min];[max]
+               </literallayout>
+
+               <para>Notes:</para>
+               <orderedlist>
+               <listitem><para>space separated list of label/value pairs</para>
+                       </listitem>
+               <listitem><para>label can contain any characters</para>
+                       </listitem>
+               <listitem><para>the single quotes for the label are optional. Required if 
+                       spaces, = or ' are in the label</para>
+                       </listitem>
+               <listitem><para>label length is arbitrary, but ideally the first 19 characters
+                       are unique (due to a limitation in RRD). Be aware of a limitation in the
+                       amount of data that NRPE returns to Nagios</para>
+                       </listitem>
+               <listitem><para>to specify a quote character, use two single quotes</para>
+                       </listitem>
+               <listitem><para>warn, crit, min or max may be null (for example, if the threshold is 
+                       not defined or min and max do not apply). Trailing unfilled semicolons can be
+                       dropped</para>
+                       </listitem>
+               <listitem><para>min and max are not required if UOM=%</para>
+                       </listitem>
+               <listitem><para>value, min and max in class [-0-9.]. Must all be the
+                       same UOM</para>
+                       </listitem>
+               <listitem><para>warn and crit are in the range format (see 
+                       <xref linkend="thresholdformat">). Must be the same UOM</para>
+                       </listitem>
+               <listitem><para>UOM (unit of measurement) is one of:</para>
+                       <orderedlist>
+                       <listitem><para>no unit specified - assume a number (int or float) 
+                               of things (eg, users, processes, load averages)</para>
+                               </listitem>
+                       <listitem><para>s - seconds (also us, ms)</para></listitem>
+                       <listitem><para>% - percentage</para></listitem>
+                       <listitem><para>B - bytes (also KB, MB, TB)</para></listitem>
+                       <listitem><para>c - a continous counter (such as bytes
+                               transmitted on an interface)</para></listitem>
+                       </orderedlist>
+                       </listitem>
+               </orderedlist>
+
+               <para>It is up to third party programs to convert the Nagios plugins 
+               performance data into graphs.</para>
+               </section>
+
+       <section><title>Translations</title>
+       <para>If possible, use translation tools for all output. Currently, most of the core C plugins 
+       use gettext for translation. General guidelines are:</para>
 
+       <orderedlist>
+       <listitem><para>short help is not translated</para></listitem>
+       <listitem><para>long help has options in English language, but text translated</para></listitem>
+       <listitem><para>"Copyright" kept in English</para></listitem>
+       <listitem><para>copyright holder names kept in original text</para></listitem>
+       </orderedlist>
+       </section>
 </section>
 
 <section id="SysCmdAuxFiles"><title>System Commands and Auxiliary Files</title>
                <orderedlist>
                        
                        <listitem><para> Do not use BEGIN and END blocks since they will be called 
-                       the first time and when Nagios shuts down with Embedded Perl (ePN).  In 
+                       only once (when Nagios starts and shuts down) with Embedded Perl (ePN).  In 
                        particular, do not use BEGIN blocks to initialize variables.</para>
                        </listitem>
          
                        <listitem><para>To use utils.pm, you need to provide a full path to the
-                       module in order for it to work with ePN.</para>
+                       module in order for it to work.</para>
                        
          <literallayout>
          e.g.
                        variable. </para>
                        
 
-                       <para>Explicitly initialize each varialable in use.  Otherwise with
-                       caching enabled, the plugin will not be recompilied each time, and
+                       <para>Explicitly initialize each variable in use.  Otherwise with
+                       caching enabled, the plugin will not be recompiled each time, and
                        therefore Perl will not reinitialize all the variables.  All old
                        variable values will still be in effect.</para>
                        </listitem>
                        
-                       <listitem><para>Do not use < DATA > (these simply do not compile under ePN).</para>
+                       <listitem><para>Do not use &gt;DATA&lt; handles (these simply do not compile under ePN).</para>
                        </listitem>
 
-                       <listitem><para>Do not use named subroutines</para> 
+                       <listitem><para>Do not use global variables in named subroutines. This is bad practise anyway, but with ePN the
+                       compiler will report an error "&lt;global_var&gt; will not stay shared ..". Values used by
+                       subroutines should be passed in the argument list.</para> 
                        </listitem>
 
                        <listitem><para>If writing to a file (perhaps recording
                        performance data) explicitly close close it.  The plugin never
-                       calls <emphasis role=strong>exit</emphasis>; that is caught by
+                       calls <emphasis role="strong">exit</emphasis>; that is caught by
                        p1.pl, so output streams are never closed.</para>
                        </listitem>
                
                        <listitem><para>As in <xref linkend="runtime"> all plugins need 
                        to monitor their runtime, specially if they are using network
-                       resources.  Use of the <emphasis>alarm</emphasis> is recommended.
+                       resources.  Use of the <emphasis>alarm</emphasis> is recommended
+                       noting that some Perl modules (eg LWP) manage timers, so that an alarm
+                       set by a plugin using such a module is overwritten by the module.
+                       (workarounds are cunning (TM) or using the module timer)
                        Plugins may import a default time out ($TIMEOUT) from utils.pm.
                        </para>
                        </listitem>
 
                <para>The option -v or --verbose should be present in all plugins.
                The user should be allowed to specify -v multiple times to increase
-               the verbosity level, as described in <xref linkend="verbose_levels">.</para>
+               the verbosity level, as described in <xref linkend="verboselevels">.</para>
     </section>
 
     <section>
        <section><title>Comments</title>
        <para>You should use /* */ for comments and not // as some compilers
        do not handle the latter form.</para>
+       <para>If you have copied a routine from another source, make sure the licence
+       from your source allows this. Add a comment referencing the ACKNOWLEDGEMENTS
+       file, where you can put more detail about the source.</para>
+       <para>For contributed code, do not add any named credits in the source code 
+       - contributors should be added into the THANKS.in file instead. 
+       </para>
        </section>
 
        <section><title>CVS comments</title>
          </varlistentry>
        </variablelist>
        </para>
+       <para>If the change is due to a contribution, please quote the contributor's name 
+       and, if applicable, add the SourceForge Tracker number. Don't forget to 
+update the THANKS.in file.</para>
        </section>
 </section>
 
-<section id="SubmittingChanges"><title>New submissions and patches</title>
+<section id="SubmittingChanges"><title>Submission of new plugins and patches</title>
+
+       <section id="Patches"><title>Patches</title>
+       <para>If you have a bug patch, please supply a unified or context diff against the
+       version you are using. For new features, please supply a diff against
+       the CVS HEAD version.</para>
+
+       <para>Patches should be submitted via 
+       <ulink url="http://sourceforge.net/tracker/?group_id=29880&amp;atid=397599">SourceForge's
+       tracker system for Nagiosplug patches</ulink> 
+       and be announced to the nagiosplug-devel mailing list.</para>
+
+       <para>Submission of a patch implies that the submmitter acknowledges that they
+       are the author of the code (or have permission from the author to release the code)
+       and agree that the code can be released under the GPL. The copyright for the changes will 
+       then revert to the Nagios Plugin Development Team - this is required so that any copyright 
+       infringements can be investigated quickly without contacting a huge list of copyright holders.
+       Credit will always be given for any patches through a THANKS file in the distribution.</para>
+       </section>
 
+       <section id="Newplugins"><title>New plugins</title>
        <para>If you would like others to use your plugins and have it included in
        the standard distribution, please include patches for the relevant
        configuration files, in particular "configure.in". Otherwise submitted 
@@ -550,8 +667,10 @@ utils (perl or c or sh) rather than cooking it's own</para>
        </listitem>
       </orderedlist>
 
-       <para>Patches should be submitted via the SourceForge and be announced to
-       the mailing list.</para>
+       <para>New plugins should be submitted via 
+       <ulink url="http://sourceforge.net/tracker/?group_id=29880&amp;atid=541465">SourceForge's
+       tracker system for Nagiosplug new plugins</ulink> 
+       and be announced to the nagiosplug-devel mailing list.</para>
        
        <para>For new plugins, provide a diff to add to the EXTRAS list (configure.in) 
        unless you are fairly sure that the plugin will work for all platforms with 
@@ -559,7 +678,59 @@ utils (perl or c or sh) rather than cooking it's own</para>
 
        <para>If possible please submit a test harness. Documentation on sample
        tests coming soon.</para>
+       </section>
+
+</section>
 
+<section id="UsingSourceforge"><title>Using Sourceforge</title>
+<table id="Projectmemberaccess"><title>Project Member Access</title>
+       <tgroup cols="9">
+               <thead>
+                       <row>
+                               <entry><para>Member type</para></entry>
+                               <entry><para>CVS Access</para></entry>
+                               <entry><para>Shell Access</para></entry>
+                               <entry><para>Release Tech</para></entry>
+                               <entry><para>Tracker Manager</para></entry>
+                               <entry><para>Task Manager</para></entry>
+                               <entry><para>Forums</para></entry>
+                               <entry><para>Doc Manager</para></entry>
+                               <entry><para>News</para></entry>
+                               <entry><para>Screenshots</para></entry>
+                               <entry><para>Notes</para></entry>
+                       </row>
+               </thead>
+               <tbody>
+                       <row>
+                               <entry><para>Translator</para></entry>
+                               <entry><para>Yes</para></entry>
+                               <entry><para>Yes</para></entry>
+                               <entry><para>No</para></entry>
+                               <entry><para>-</para></entry>
+                               <entry><para>-</para></entry>
+                               <entry><para>-</para></entry>
+                               <entry><para>-</para></entry>
+                               <entry><para>-</para></entry>
+                               <entry><para>-</para></entry>
+                               <entry><para>Add developer to CVSROOT/avail for translation files</para></entry>
+                       </row>
+                       <row>
+                               <entry><para>Developer</para></entry>
+                               <entry><para>Yes</para></entry>
+                               <entry><para>Yes</para></entry>
+                               <entry><para>No</para></entry>
+                               <entry><para>-</para></entry>
+                               <entry><para>A&amp;T</para></entry>
+                               <entry><para>Moderator</para></entry>
+                               <entry><para>Editor</para></entry>
+                               <entry><para>-</para></entry>
+                               <entry><para>-</para></entry>
+                               <entry><para>Need to set tracker access individually. Add developer to CVSROOT/avail for all files</para></entry>
+                       </row>
+               </tbody>
+       </tgroup>
+</table>
+<para>Add all members to the nagiosplug-team mailing list</para>
 </section>
 </article>