====== ndo2fs  ======

**WORK IN PROGRESS**

ndo2fs originated as a debugger for ndomod.o. It takes data from ndomod.o either via a TCP or a UNIX socket.
The aim of ndo2fs is to save the nagios data in a structured way using the file system. We kept in mind that the data has to be processed in an easy way.

In addition I'm working on a PHP class to access the data of ndo2fs. For more information read [[ndo2fs-php]].

===== Mode of operation =====

ndo2fs works like ndo2db but in contrast to ndo2db it is written in Perl and saves the data on the file system.
That's about all.









===== Download =====

The current version can be found in the GIT Repository under [[http://git.nagiosprojects.org/?p=ndo2fs.git|git.nagiosprojects.org]].

<box 90% |**Current GIT Repository Snapshots:**>
\\
[[http://git.nagiosprojects.org/?p=ndo2fs.git;a=snapshot;h=0.2.1|GIT Snapshot ndo2fs-0.2.1]]\\
[[http://git.nagiosprojects.org/?p=ndo2fs.git;a=snapshot;h=refs/heads/master|GIT Snapshot Head]]\\
</box>

\\

The [[http://www.nagvis.org|NagVis]] developers Lars Michelsen and Michael Lübben as well as the [[http://www.my-plugin.de/wiki/de/projects/check_multi/start|check_multi]] developer Matthias Flacke provide valuable input right now.

In case of questions please send a mail to [[pitchfork@pnp4nagios.org]]



===== Installation =====

None yet! It's just a single Perl script.

Since version 0.1.13 the Perl module JSON::XS is needed to create the JSON files. Unfortunately the module is available as a [[http://packages.debian.org/de/lenny/libjson-xs-perl|package]] starting with Debian Lenny.
Users of Debian Etch have to install it via CPAN typing
<code>perl -MCPAN -e install JSON::XS</code>

===== ndomod data_processing_options =====

The data processed by ndomod can be minimized using the directive "data_processing_options" defined in ndomod.cfg as ndo2fs will only use a part of the data.

<code>
              PROCESS_PROCESS_DATA  1
        PROCESS_SERVICE_CHECK_DATA  64
           PROCESS_HOST_CHECK_DATA  128
              PROCESS_COMMENT_DATA  256
             PROCESS_DOWNTIME_DATA  512
       PROCESS_PROGRAM_STATUS_DATA  2048
          PROCESS_HOST_STATUS_DATA  4096
       PROCESS_SERVICE_STATUS_DATA  8192
        PROCESS_OBJECT_CONFIG_DATA  262144
          PROCESS_MAIN_CONFIG_DATA  524288
            PROCESS_RETENTION_DATA  2097152
      PROCESS_ACKNOWLEDGEMENT_DATA  4194304
          PROCESS_STATECHANGE_DATA  8388608
       PROCESS_CONTACT_STATUS_DATA  16777216
</code>
which adds up to
<code>
          data_processing_options = 32259009
</code>

===== TODO =====

  * Daemon Code
  * Checks
  * Code cleanup

===== Performance =====

1.4m lines equate to 10 min on my production system

   wc -l ndo.dat 
   1406535 ndo.dat

This data I passed to ndo2fs via NetCat

<code>
time cat ndo.dat | nc -q1 -n 127.0.0.1 5668

real	0m19.561s
user	0m0.004s
sys	0m0.064s
</code>

This results in 67 MB of data in the ndo2fs directory

<code>
du -s -h /tmp/ndo2fs
67M	/tmp/ndo2fs
</code>
 
The measurements were taken from a Debian 4 VM on my Macbook Pro.

===== Data format =====

The data is stored in JSON format on the file system. JSON is short for "[[http://www.json.org|JavaScript Object Notation]]" but don't panic, there are enough easy to use parsers.

Example using PHP5:

<code php>
<?php
$in = '/tmp/ndo2fs/VOLATILE/HOSTS/localhost/http/STATUS';
$data = file_get_contents($in);
$out = json_decode($data);
var_dump($out);
?>
</code>

===== Data structure =====

Taking ''/tmp/ndo2fs'' as a base directory you'll find two sub directories after the start of ndo2fs.

  * ndo2fs
    * VOLATILE
    * PERSISTENT

The directory VOLATILE contains (what a surprise) volatile data which is created at the start of nagios. That means that VOLATILE can be stored on a TMPFS.
PERSISTENT contains data which is created during runtime of nagios and will be kept across restarts.

The nagios instance forms the next level. For each nagios instance there will be a directory with the appropriate name. In the following the instance <default> will be used.

==== VOLATILE ====

Below of VOLATILE/default/ there will be the data of the instance <default>.

  * **default**
    * **HOSTS**           # directory for host/service state and config
      * **<hostname>**    # directory per host
        * //CONFIG//      # config  
        * //STATUS//      # current state
        * **<servicedesc>**  # directory per service
          * //CONFIG//    # config
          * //STATUS//    # state
    * **HOSTGROUPS**
      * **<hostgroupname>**
        * //META//        # meta information regarding the hostgroup
    * **SERVICEGROUPS**
      * **<servicegroupname>**
        * //META//
    * **CONTACTGROUPS**
      * **<contactgroupname>**
        * //META//         # meta information regarding the contactgroup
        * __<contact>__    # a symlink to the appropriate contact for each member of the group
    * **CONTACTS**
      * //<contactname>//  # one file per contact
    * **HOSTAUTH**
      * //<contact>//  # Array of services a contact is authorized for. 
    * **SERVICEAUTH**
      * //<contact>//  # Array of host and services a contact is authorized for. 

    * **HOSTSTATE**
      * **<state>**   # directory per state ( UP,DOWN,UNREACHABLE )
        * **<hostname>**
          * __STATUS__  # symlink on the STATE file of the host
    * **SERVICESTATE**
      * **<state>**   # directory per state ( OK,WARNING,CRITICAL,UNKNOWN )
        * **<hostname>**
          * **<servicedesc>**
             * __STATUS__  # symlink on the STATE file of the service
    * **VIEWS**
      * //HOSTLIST//  # list of all hosts
      * //SERVICELIST//  # list of all services
      * //PARENTLIST//  # list of the parent relations between hosts 

==== PERSISTENT ====

Below of PERSISTENT/default/ there will be data which is collected during nagios runtime and kept across restarts

  * **default**
    * **DOWNTIMES**           # directory for each host/service state and config
      * //<downtimeid>//  # each downtime in a separate file. The file name is the downtime ID
    * **COMMENTS**
      * //<commentid>//  # each comment in a separate file. The file name is the comment ID
      * //<host>::<service>//  # each comment in a separate file. The file name consists of host name and service description
    * **LOGS** 
      * //YYYYMMDD.log// # One logfile per day.