Nagios Business Process View and Nagios Business Impact Analysis
----------------------------------------------------------------
The software and documents in this package have been produced by
Sparda-Datenverarbeitung eG, Nuernberg, Germany, Bernd Stroessenreuther
<berny1@users.sourceforge.net> and are available to the community under the
conditions of GNU General Public License Version 2, see LICENSE.
Short overview
--------------
The AddOn "Business Process View" takes results of the single nagios checks out
of NDO or IDO backend (NDO database, IDO database, ndo2fs, Merlin, mk_livestatus,
Icinga-API) and builds up aggregated states.
How they are associated is described in one or more config files.
There is the possibility to make "and" conjuctions, "or" conjunction and other...
A business process (as defined by such a formula) can be used as a part of
another business process. So You can build up a hirachical structure to describe
the state of Your Application.
The AddOn "Business Impact Analysis" allows You to simulate Outages. You can set
manually the state of each single component to each state You like and look, how
this would impact Your applications.
Help
----
If You have problems installing or using this AddOns, please visit the
Support page on our homepage:
http://nagiosbp.projects.nagiosforge.org/support.shtml
Here You find a FAQ and some helpful mailinglists.
Also a very active community of users You find at
http://www.nagios-portal.org/
(this one is in german only)
What is it?
-----------
(a little bit more in detail)
You are running a lot of applications for Your customers. (I use the word
customers because as a system administrator You allways have customers,
no matter if they are employees or customers of the company You are working
for.)
Each application needs a few or a lot of components (like webservers,
application servers, DNS- or mail servers, LAN- or WAN-connections ...)
to work properly.
There are components You need for only one application, and of course there
are components which are important for more applications (e. g. DNS-servers)
You already are running Nagios or Icinga to monitor all of these components
I guess. (Otherwise You would not think about this AddOn.)
If You are the only system administrator of Your company, You will probably
know all Your applications very well, You know which application needs
which components - then You will not need this AddOn.
If there are more admins, You probably will share work. This means each admin
knows view applications very well and the other applications only a little
bit.
So maybe You would find it great to visualize, how all these components work
together. If one ore more components fail, You want to see on one single
page, which applications are unavailable for Your customers - in this case:
Install this addon.
It has two modes:
1. Nagios Business Process View
it shows the actual state of Your applications
2. Nagios Business Impact Analysis
this is a simulation mode.
You can set each of Your components to every stat You like.
So if You want to know: What would be if my web server would fail now?
Just klick the state of Your web server an set it to CRITICAL
Return to the overview page and look, which applications are now in
state CRITICAL.
The states of the single services and hosts defined in Nagios or Icinga are
taken from the NDO or IDO backend (NDO database, IDO database, ndo2fs, Merlin,
mk_livestatus or Icinga-API).
How it works
------------
You have one or more config files in which You define Your applications.
You define which components are needed and how they are related.
So go and set up a config file called etc/nagios-bp.conf
There You have to type some simple formulas for defining business processes.
e. g.
loadbalancers = loadbalancer1;System Health | loadbalancer2;System Health
website_webserver1 = webserver1;HTTP & webserver1;HTTPD Slots
The first string is the name You want to give to the business process.
On the right side You have strings in the form
<hostname>;<servicename>
The example above means:
You have a loadbalancer cluster. If one of them is in ok state, the
application is available for the customer. So You define a "or" conjuction
for Your business process.
If You are looking if Your webserver1 works well, You normaly look for the
Check HTTP and also for the check "HTTPD Slots". If both are in OK state, You
know, the webserver1 is working well.
So we put these two together by making a "and" conjuction.
Next step is, to give a name to each business process You defined, so type
display 0;loadbalancers;Loadbalancer Cluster
display 0;website_webserver1;WebServer 1
The digit after the keyword display is the priority class, in which these
business process ist displayed in the top level view.
0 means: No display
1, 2,...: Display in the given priority.
As You can use single business processes again in other processes, display 0
is very useful, if You do not want to display each sub-process in the top
level view.
Let's have a complete example:
internetconnection = internetconnection;Provider 1 | internetconnection;Provider 2
display 0;internetconnection;Internet Connection
loadbalancers = loadbalancer1;System Health | loadbalancer2;System Health
display 0;loadbalancers;Loadbalancer Cluster
dns = dns1;DNS | dns2;DNS | dns3;DNS
display 0;dns;DNS Cluster
website_webserver1 = webserver1;HTTP & webserver1;HTTPD Slots
website_webserver2 = webserver2;HTTP & webserver2;HTTPD Slots
website_webservers = website_webserver1 | website_webserver2
website = internetconnection & loadbalancers & dns & website_webservers
display 0;website_webserver1;WebServer 1
display 0;website_webserver2;WebServer 2
display 0;website_webservers;WebServer Cluster
display 1;website;WebSite
If these line are the only ones in Your nagios-bp.conf file, this should work.
You have defined Your first business process! Congratulations!
Go and view http://your-host/nagiosbp/cgi-bin/nagios-bp.cgi
(I just assume, You have all these services and hosts defined in your Nagios
or Icinga configuration or adapted the example.)
Care for the correct spelling! The <hostname> and the <servicename> must exactly
match the spelling, how You defined them in Nagios/Icinga. Watch for correct upper
case and lower case.
More examples can be found in etc/nagios-bp.conf
Syntax check
------------
To make sure the syntax of Your nagios-bp.conf is correct, run
bin/nagios-bp-consistency-check.pl
In this syntax it checks Your default nagios-bp.conf. If You want to check
some other file call
bin/nagios-bp-consistency-check.pl <file>
Some more keywords
------------------
In the top level view (http://your-host/nagiosbp/cgi-bin/nagios-bp.cgi)
the right column is empty at the moment. It can be used to display some short
information according the business process. This can be a static or dynamic
string. e. g. You want to display, how many users are currently logged into
Your webshop if You defined a business process WebShop.
Or You want to display a short announcement, ...
Just write a little script that displays the information You like (just one
line to stdout) and the You configure:
external_info website;echo '<b>Please note:</b> Today maintainance on WebServer1,<br>Production only on WebServer2'
or
external_info website;/path/to/your/script.sh
Maybe one little string is not enough for all the information You have.
Then
info_url website;/more_info/website.html
or
info_url website;http://some.other.site.com/more_info/website.html
would be of value for You. Just linking to a WebSite with all the information.
In the first syntax, the page is located on the Nagios/Icinga machine.
In the second syntax, it is somewhere in the world.
If You defined some info_url, a little info icon appears, which can be clicked
by Your users.
Maybe You want to use it for some emergency documentation or so.
complete syntax description
---------------------------
PLEASE NOTE: The order of Your definitions is important!!
If You use a (sub level) business process in the definition of another business
process, make sure You define the subl level process BEFORE You use it.
<bp_name> = <host>;<service> [& <host>;<service>]+
Services have a "and" conjuction. All of them are needed for the application to
be available to the customer.
Or in other words: If all of the given services are OK, the defined business
process has state OK. If at least one is WARNING, the process has state
WARNING. If at least one is CRITICAL, the process gets CRITICAL.
<bp_name> = <host>;<service> [| <host>;<service>]+
Services have a "or" conjunction. This is often used if You have redundant
systems. If one of them is working, the application is available to the
customer.
If at least one service is OK, the process gets state OK.
If all services are CRITIAL and at least one is WARNING, the process gets
state WARNING.
Only if all of the services are CRITIAL, the process gets CRITICAL.
<bp_name> = <min_num> of: <host>;<service> + <host>;<service> [+ <host>;<service>]+
Use this one, if You have a number of application servers running the same
application and You know You need at least <min_num> of servers active for
load reasons.
e. g.
appserver_cluster = 2 of: appserver1;WebShop + appserver2;WebShop + appserver3;WebShop + appserver4;WebShop
So if at least 2 of the given services are in state OK, the process is OK.
<host>;Hoststatus
You also can use the results of Nagios/Icinga host checks in Your business
processes. In this case You use this syntax.
Instead of <host>;<service> You always can use <bp_name> too, where <bp_name>
is the name of a business process You defined BEFORE.
display <x>;<bp_name>;<long_name>
The digit x is the priority of the process. The process is displayed in this
priority class in the top level view.
0 means: This process is not displayed in the top level view.
<long_name> is the name or description used when displaying the process.
(The user never sees <bp_name> in the GUI, always <long_name>)
external_info <bp_name>;<script>
For each business process which is not defined with display 0 you can add a
script. The script must print one line to stdout. This line is displayed
in the right column of the top level view near the business process.
info_url <bp_name>;<url>
For each business process which is not defined with display 0 you can add a
URL. If You define one, a little info icon (white letter i on blue ground)
is displayed near the business process. Clicking this icon brings You to the
given URL.
template <bp_name>;<service_template>
This is only used, if You map Business Processes as services in Nagios or
Icinga. (see below "Business Process representation as Nagios/Icinga services")
Normaly all Business Processes with display greater than 0 get the same
service template and all Business Processes with display 0 get another one.
If You want to have one Business Process e. g. with a different
notification_period, You just define one more template and give it's name
here.
Check if everything works
-------------------------
If You did everything described until this point and set up as described in
INSTALL, You probably have got a nice new view in Your Nagios/Icinga installation.
If not, DO NOT STEP BEYOND THIS POINT.
I recommend, You go and solve Your problems first, get Your installation running
and then continue with the next (advanced) part.
If the page "Business Process View" is not displayed, check the error log
of Your webserver for the problem. Maybe a required perl module is not found
or You have a problem with file permissions.
If You have SELinux or AppArmor, maybe they do not allow access to some
files.
Maybe You just forgot to restart Your webserver after making the config changes.
Next thing to do to locate the problem is to run a script especially built
for this:
bin/nagios-bp-check-ndo-connection.pl
Run it under the user Your Apache runs with. On success it prints out a
list of all the services Your NDO or IDO backend knows together with status
information and the last plugin output.
On error You should get an error message that probably will be very helpful
for locating the problem.
To see if the problem is in the webserver or in Nagios Business Process AddOn
You can call
sbin/nagios-bp.cgi
form the commandline. Do this as a normal user, not as root.
It should print out some HTML code on STDOUT, if everything works and
error or warning messages otherwise.
Business Process representation as Nagios/Icinga services
---------------------------------------------------------
You now have a beautiful page that displays the health of all or some of Your
applications at any time.
But You always get this information for NOW. You do not see, how it was an
hour ago or one day ago. You have no history.
And: You do not get notifications for whole business processes, only for single
components.
Solution is simple. Just make any business process itself to be a normal Nagios
or Icinga service. Nagios checks the state of these services regularly, e. g. each
minute. On these service You can use all of Nagios' or Icingas features:
notifications, statistics, history, ...
A simple script is integrated, that generates an additional Nagios/Icinga
configuration file with all the business processes as service.
The script can be found under
bin/bp_cfg2service_cfg.pl
Running this script should produce a new file
services-bp.cfg
in the etc directory of NAGIOS/ICINGA (not etc of Nagios Business Process AddOns)
Do not make any manual changes to this file! If You change Your business
processes later You just run this script again. This produces an actual
services-bp.cfg but would overwrite any changes You made there.
Next step is to edit the nagios.cfg or icinga.cfg file (normally
/usr/local/nagios/etc/nagios.cfg or /usr/local/icinga/etc/icinga.cfg)
and add the new config file by adding a line like:
cfg_file=/usr/local/nagios/etc/services-bp.cfg
or
cfg_file=/usr/local/icinga/etc/services-bp.cfg
Additionaly You need two new service templates, two new dummy hosts, and two
new commands looking like the following.
Please define them in another Nagios/Icinga configfile, e. g. the service in
services.cfg, the hosts in host.cfg and the commands in commands.cfg
(Do not define them in services-bp.cfg!!)
define service{
name generic-bp-service
use generic-service
contact_groups nagios-admins
host_name business_processes
notification_period none
max_check_attempts 1
register 0
}
define service{
name generic-bp-detail-service
use generic-service
contact_groups nagios-admins
host_name business_processes_detail
notification_period none
max_check_attempts 1
normal_check_interval 5
retry_check_interval 1
register 0
}
define host{
use generic-host
host_name business_processes
alias Business Processe
address 10.6.255.99 # dummy IP
contact_groups nogroup
check_command return_true
}
define host{
use generic-host
host_name business_processes_detail
alias untergeordnete Business Processe
address 10.6.255.99 # dummy IP
contact_groups nogroup
check_command return_true
}
define command{
command_name check_bp_status
command_line /usr/local/nagiosbp/libexec/check_bp_status.pl -b $ARG1$ -f $ARG2$
}
define command{
command_name return_true
command_line $USER1$/check_dummy 0
}
Now reload Your Nagios/Icinga configuration. That's it.
Nagios/Icinga now checks the states of all business processes regulary.
If only states of business processes in the top level view are of interest to You,
but not the states of the sub-processes, use
bin/bp_cfg2service_cfg.pl -s 0
In this case You do not need the service template generic-bp-detail-service and
the host template business_processes_detail
Parameters for this script are displayed, if You type
bin/bp_cfg2service_cfg.pl --help
If You are using these new services to generate alerts for problems of whole
business processes, You might not want to inform the same users or groups for
all business processes. Of course this would be the case, if You have only one
template for all of them (as described above).
So for all business processes that should use some other template than the
default, You just enter a line in the form
template <bp_name>;<service_template>
e. g. for our example
template website;website-bp-service
to Your etc/nagios-bp.conf.
This means, the service representing the business process website should be
generated with the template website-bp-service.
Of course, in this case You need to define an other template:
define service{
name website-bp-service
use generic-bp-service
contact_groups website-admins
notification_period 24x7
register 0
}
Run bin/bp_cfg2service_cfg.pl again. It will produce a services-bp.cfg where
the according service is now defined referring to this new template.
Reload Nagios/Icinga and You are done.
More than one Top Level View
----------------------------
Until now, all the business processes You defined show up in the one and only
top level view. Maybe You want to have more than one top level view, e. g.
because You have different customers and each of them should only see his
business processes.
So You just go to etc directory and build additional configuration
files, each of them defining business processes as described above.
Let's assume You called them
nagios-bp-customer1.conf
nagios-bp-customer2.conf
nagios-bp-customer3.conf
...
they all must be located in the same directory as nagios-bp.conf
Now use the following URLs to view them:
For the business process view:
http://<servername>/nagiosbp/cgi-bin/nagios-bp.cgi?conf=nagios-bp-customer1
http://<servername>/nagiosbp/cgi-bin/nagios-bp.cgi?conf=nagios-bp-customer2
http://<servername>/nagiosbp/cgi-bin/nagios-bp.cgi?conf=nagios-bp-customer3
...
And for the business impact analysis:
http://<servername>/nagiosbp/cgi-bin/nagios-bp.cgi?conf=nagios-bp-customer1&mode=bi
http://<servername>/nagiosbp/cgi-bin/nagios-bp.cgi?conf=nagios-bp-customer2&mode=bi
http://<servername>/nagiosbp/cgi-bin/nagios-bp.cgi?conf=nagios-bp-customer3&mode=bi
...
Please note: The different configurations do not know each other. So it is not
possible to use a business process defined in one of them in another configuration.
If You want to map the Business Processes of all of these files to Nagios
services, use
/usr/local/nagiosbp/bin/bp_cfg2service_cfg.pl -f /usr/local/nagiosbp/etc/nagios-bp-customer1.conf
/usr/local/nagiosbp/bin/bp_cfg2service_cfg.pl -f /usr/local/nagiosbp/etc/nagios-bp-customer2.conf
/usr/local/nagiosbp/bin/bp_cfg2service_cfg.pl -f /usr/local/nagiosbp/etc/nagios-bp-customer3.conf
Maybe You want different default templates for each of these config files,
because in the one case contacts of customer1 should be notified, in the
other case contacts of customer2, ...
For this You do not need to include template directives for every single
Business Process in every single config file. Just call bp_cfg2service_cfg.pl
with -tt or -tm parameters.
Complete description see
bin/bp_cfg2service_cfg.pl --help
Where used?
-----------
(new feature since version 0.9.4)
Imagine You want to stop one service or one host in Your network for maintainance.
But You are not the one who did the setup of this host or service and You are
not sure in which of Your business processes it is used.
Ok, You could use the webinterface and click on every business process and
look if it is listed there. Quite a lot of time if You have lots of business
processes. So You might want to use this feature:
You can use Nagios' or Icingas notes_url or the action_url to put the whereUsed-URL
into:
define host {
# ...
notes_url /nagiosbp/cgi-bin/whereUsed.cgi?host=$HOSTNAME$
}
or
define host {
# ...
action_url /nagiosbp/cgi-bin/whereUsed.cgi?host=$HOSTNAME$
}
at services:
define service {
# ...
notes_url /nagiosbp/cgi-bin/whereUsed.cgi?host=$HOSTNAME$&service=$SERVICEDESC$
}
or
define service {
# ...
action_url /nagiosbp/cgi-bin/whereUsed.cgi?host=$HOSTNAME$&service=$SERVICEDESC$
}
These definitions work for Nagios 3.x and Icinga. When still using Nagios 2.x,
please note, that You have to define notes_url or action_url directives in
hostextinfo or serviceextinfo sections, not directly in the host or service
section.
Probably it will be the best choice to define these notes_url or action_url
in a template, e. g. in generic-host or generic-service to have them available
for all Your services and hosts.
With these definitions You get new icons in Your Nagios' or Icingas
webinterface for every host and service. When clicking it You are presented a
new view showing in which of Your business processes the according host or
service is used. Here You are!
If Your notes_url and action_url both are already used for other important
things like PnP and other AddOns, You still have another (second best)
chance. You can define a link in the text You are giving as "notes".
(again please note, this directive has to be defined in hostextinfo or
serviceextinfo section in Nagios 2.x)
define host {
# ...
notes On problems with this host, see documentation chapter 16.<br><a href="/nagiosbp/cgi-bin/whereUsed.cgi">Where is this host used?</a>
}
define service {
# ...
notes On problems with this service, see documentation chapter 16.<br><a href="/nagiosbp/cgi-bin/whereUsed.cgi">Where is this service used?</a>
}
The first part is the normal information for the host or service. You maybe
did define this part already to give the users some helpful information in
the webinterface.
Next You see a <br> for a new line and then a link in normal HTML syntax.
You may have noticed, we do not add the parameters host=$HOSTNAME$ and
service=$SERVICEDESC$ as we did above. This is because Nagios/Icinga does
not expand macros here. So we do not have this chance.
Instead the whereUsed.cgi looks for the HTTP referer (a HTTP request
header telling which URL the click did come from) and tries to expand
the information about host and/or service from this.
This only works, when using the official Nagios/Icinga webinterface. When
using some other GUI, it will probably not work.
Some users also configure their browsers to suppress HTTP referers (to
increase privacy) e. g. by using a special Firefox Plugin. In this case
it also will not work.
In both cases it will be the better solution to use action_url or
notes_url.
Additional hint: If 3 links (one in action_url, one in notes_url and one
in notes) are not enough for everything You want to link from one single
host or service, e. g. different AddOns, a link to Your server documentation
and so on, You can add more than one link in the text given in the notes
directive.
Where used with more than one Top Level View
--------------------------------------------
If You have more than one Top Level View (see according section above) and
want to use the "Where used?" feature, by default the whereUsed.cgi looks for
processes in the default config file nagios-bp.conf. If You want it to look
in an alternate config file, call it with the additional parameter "conf"
in the same way as described in section "More than one Top Level View".
e. g.
action_url /nagiosbp/cgi-bin/whereUsed.cgi?conf=nagios-bp-customer1&host=$HOSTNAME$&service=$SERVICEDESC$
or
notes <a href="/nagiosbp/cgi-bin/whereUsed.cgi?conf=nagios-bp-customer1">Where is this service used?</a>
Preview Changes
---------------
If You make changes to Your nagios-bp.conf file, the new configuration is used
as soon as You save the file. All Your customers see Your changes immediatelly.
Maybe this is not what You want. Maybe You want to check if everything is how
You desinged it and there are no errors in the file.
So copy nagios-bp.conf to e. g. nagios-bp-new.conf and make Your changes to
this copy. Now it's a good idea to make a syntax check of nagios-bp-new.conf
as described in section Syntax check.
Now You can preview as described in section More than one Top Level View
If everything is ok, You copy it over to nagios-bp.conf to bring Your changes
online.
Using new hosts or service in a Business Process
------------------------------------------------
If You just did define new hosts or services in Nagios/Icinga and You want to
use them in a Business Process, make sure You did reload Your Nagios/Icinga
first and then You wait a minute. This gives Nagios/Icinga the chance to write
the first status information to the NDO or IDO backend (NDO database, IDO
database, ndo2fs, Merlin, mk_livestatus, ...). Now You can use them.
Also the syntax check (see above, section Syntax Check) can only work if
the first status information is available in the NDO or IDO backend.
Caching results from NDO or IDO backend
---------------------------------------
(new feature since version 0.9.4)
If You have performance problems with the NDO or IDO backend You are using,
this might result in too high latency of Your Nagios/Icinga installation.
Best thing in this case is to tune Your NDO or IDO backend, e. g. by tuning
MySQL or You decide to write as less information as possible into the NDO or
IDO backend.
Only if this is impossible, or does not make Your system as fast as required
You might think about caching values in the Nagios Business Process AddOn for
a certain amount of time. So access to Your NDO or IDO backend is done not so
often, but the results are not so fresh as they could be.
Cache is held in one file in the filesystem, by default
var/cache/ndo_backend_cache
If You did really decide to use caching, You activate it by using the
parameters cache_time and cache_file in etc/ndo.cfg
(they have comments there)
Please note: var/cache/ndo_backend_cache is created world-writable
(permissions 666) by the installer. If You do not want this for security
reasons, You can change permissions to whatever You like. The only thing You
must make sure is that the user Your Nagios/Icinga is running under and the
user Your webserver is running under have write access to this file.
Adapting the GUI to fit Your needs
----------------------------------
(new feature since version 0.9.5)
Some users did ask for some changes in the GUI.
The difficulty is, the AddOn is used for so many different purposes, so
that it is nearly impossible to build a GUI which fits for all of them.
Some use it for display on a big monitoring screen in an IT operations center
and like big letters to be able to read it from a distance of some meters.
Others use it in the browser on a smart phone and need some compact view
without needing to scroll so much.
Some use many priorities, some need only one priority (so navigation
between priorities is unneccessary).
I decided to have a generic solution: By default keep the GUI as it was
because I think this layout is ok for most users.
In addition giving each installation the possibility to change the layout
of the GUI to fit special needs using an own CSS.
All pages include an additional CSS file called user.css. This one You can
change if You like. All the relevant elements in the generated HTML pages have
an ID now so that everything You can do with CSS can be done with Nagios
Business Process AddOn's pages.
user.css is located in share/stylesheets/ directory.
some examples:
If You want to make Trafficlight readable from a bigger distance, just put
the following line into user.css:
#nbp_trafficlight_yes_table td { border-top:30px solid white; padding:5px; font-size:120% }
If You do not want to display the selction buttons for single priorities:
#nbp_prio_selection { display:none; }
In the same way you can hide other elements too.
To display the traffic light on the right hand side, You need two lines:
#nbp_cental_table_box_tl_yes { margin-right:9em; margin-left:0; }
#nbp_trafficlight_yes_box { float:right; }
And I'm sure users find much more cool hacks to make the GUI fit their needs.
If You found one, which You think it is interesting for other users too,
please mail it to nagiosbp-users@lists.nagiosforge.org and we will publish it.
In the approach described above all users of one installation get the same
layout of the GUI. If You like different layouts for single users, they might
want to use an addon for their browser (like Stylish for Firefox) to add user
specific styles directly in their browser.
Maybe You like a more centralistic approach: You could provide more than one
user.css on the server. The problem is, all browsers ask for the same URL of
user.css. But You can use something like apache's mod_rewrite to decide (e. g.
by the client's IP address) to deliver the one or the other.
