The Centralized Configuration structure
that you define depends on the size and organization of your monitored
enterprise, the types of monitoring agents you want to maintain, what
kinds of updates, and how often.
Configuration load list
Any newly installed
monitoring agent or an existing agent can participate in
Centralized Configuration by using a configuration
load list. The configuration load list is an XML configuration file
that is unique to the running agent. It tells the agent how to connect
to one or more
central configuration servers and what
files to download from those servers.
Administrators maintain
and update the configuration file in this centralized repository.
New agents collect initial configuration files from this location
and periodically or on-demand contact the server to collect any updates
or changes.
Central configuration
server
The configuration load list contains
one or more ConfigServer elements that tell the agent how to connect
to a central configuration server. Central configuration
servers contain a repository of files that are served to the monitoring
agents using HTTP.
The server authenticates requests, examines
the configuration article last update time stamp (set to GMT), and,
if the client copy of the requested file is older than the server
copy, the server returns the configuration article contents to the
agent. Otherwise, it returns HTTP status 304 - Object Not Modified.
The server returns other HTTP status if an error is encountered during
article processing.
When deploying a
central configuration server, consider these factors:
- User access
- Administrators require access to maintain the configuration load list that are to be distributed.
The central configuration clients need access to the central configuration server to collect the updates. If you
already have a web server and are familiar with access and maintaining
files there, you can use any available web server to act as a central configuration server. This includes the web server
for the Agent Service Interface.
- Directory structure
- Identify a directory structure on the server and client that enable
you organize your files, to minimize any duplication on the server,
and to reduce the number of files that must be maintained.
- Keyword substitution
- Using keyword substitution in the configuration load list can
simplify the organization of configuration files.
- For example, if all of the Linux OS agents run one set of private
situations with events defined and the Windows OS agents run another
set, the @PRODUCT@ keyword can be used to direct the agents to the
correct directory and file on the central configuration server.
Place the files in the Install_dir/localconfig
directory or as specified by the IRA_SERVICE_INTERFACE_CONFIG_HOME
environment variable. For example,
common
cnfglist.xml
lz
lz_cnfglist.xml
lz_trapcnfg.xml
nt
nt_cnfglist.xml
nt_trapcnfg.xml
- The cnfglist.xml file can use the @PRODUCT@ keyword to direct
the agents to the correct files. Example:
<ConfigurationArtifact>
<ConfigServer
Name="CENTRAL-CONFIG-SERVER"
URL="http://icvr5a05.tivlab.raleigh.ibm.com/"
User="itmuser"
Password="{AES256:keyfile:a}vHBiEqmmvylNPs90Dw1AhQ==" />
<ConfigFile
Server="CENTRAL-CONFIG-SERVER"
Name="cnfglist.xml"
Path="common"
Disp="CNFGLIST"
Activate="YES" />
<ConfigFile
Server="CENTRAL-CONFIG-SERVER"
Name="@PRODUCT@_situations.xml"
Path="@PRODUCT@"
Disp="PVTSIT"
Activate="YES" />
<ConfigFile
Server="CENTRAL-CONFIG-SERVER"
Name="@PRODUCT@_trapcnfg.xml"
Path="@PRODUCT@"
Disp="TRAPCNFG"
Activate="RESTART" />
</ConfigurationArtifact>
Other keywords can be used
to create the granularity required for each agent to get the correct
files.
Web server as the central configuration server
Tivoli
monitoring agents can use existing web servers to collect configuration
files. You can use any web server that the agents can access with
HTTP or HTTPS as a central configuration server.
To use a web
server, create a user ID that has permission to access the files on
the
central configuration server and reference those
credentials in the configuration load list. The URL specification
is slightly different than when the
central configuration server is
a
monitoring agent, but that is the
only difference in the
configuration load list.
The ConfigServer element looks like this:
<ConfigServer
Name="CENTRAL-CONFIG-SERVER"
URL="http://webserver.domain.com/"
User="itmuser"
Password="{AES256:keyfile:a}vHBiEqmmvylNPs90Dw1AhQ==" />
Monitoring agent as the central configuration server
All monitoring agents
(enterprise or system monitor) that run on Windows, Linux and UNIX
platforms contain an HTTP-based Service Interface that can be used
as a central configuration server. Monitoring agents
on z/OS and i5 do not provide an HTTP Service interface, so cannot
be used as central configuration servers.
The
advantage of using a monitoring agent as the central configuration server rather than a web server, is
that you do not need to maintain a web server and you can use several
agents to form a set of cascading central configuration server to
provide some workload balancing.
- User access to the central configuration server
- Access to an agent's Service Interface is controlled using
its Access Authorization Group Profile (AAGP). The ability
to request or place files on an agent requires more security rights
than viewing metrics or historical data. By default, any valid ID
on the host computer where the agent is running can access the Agent Service Interface. Those users can view metrics,
situations, and historical data collected by the agent. However, the
default behavior gives only the Administrative ID on the host
the permission to access central configuration files through the agent's
Service Interface.
- The following user IDs are default members of the Administrative group
on the platform where the agent is running:
Administrator
root
ITMUSER
QSECOFR
You can create configuration load lists and
use the Administrative credentials to connect to the monitoring
agent that is acting as the central configuration server. The
ID does not have to exist on the client agents for them to connect
to the central configuration server.
- On Windows
systems, the agent automatically discovers all user accounts belonging
to the AAGP Administrators group; the agent adds those accounts to
the AAGP Administrative group at startup. On UNIX and Linux systems,
the agent automatically discovers all user IDs having root (superuser)
authority; the agent adds them to AAGP Administrative group at startup.
(Userid QSECOFR on i5/OS and ITMUSER on z/OS remain unchanged at this
time.) As results:
- Customers do not need to create special user account to use agent
services.
- Administrator/root ID exposure is avoided.
- Any authorized account on a system automatically gains access
to privileged agent services such as file access, AAGP customization,
and so on.
- Although user IDs can be stored in encrypted format, in most cases
you want to define a user ID on the system that will be hosting the central configuration server and add that user to the agent's
AAGP.
- The following example gives the steps for creating a user on the
Linux OS agent and granting administrative access to the central configuration server there:
- Create the user on the system: The user named "itmuser" is
created on the Linux operating system and the Linux OS agent (lz)
is the central configuration server.
- Create an AAGP.xml file in the Install_dir/localconfig directory
that adds the new ID to the Administrative group:
<AAGP>
<AAUSER>
<ID>itmuser</ID>
<ASSIGN>AD</ASSIGN>
</AAUSER>
</AAGP>
The default local configuration directory
can be changed with the IRA_SERVICE_INTERFACE_CONFIG_HOME environment
variable. Tip: Consider having each agent
collect an AAGP file so that you can contact the agent directly through
its Agent Service Interface to perform configuration
actions with a non-root ID. By connecting directly to an agent's Service
Interface with AD permission, you can provide credentials to connect
to a new central configuration server, put or get files,
or force immediate refreshes of configuration files.
- On the monitoring agent that functions as the central configuration server, edit the configuration load
list to add a DISP="AAGP" ConfigFile entry to load the AAGP XML file.
This
load list must use credentials (

root;

Administrator) to connect to itself to add the new user ID
to the AAGP. The edited
configuration load list looks
like this:
<ConfigurationArtifact>
<ConfigServer
Name="CENTRAL-CONFIG-SERVER"
URL="http://linuxhost:1920///linuxhost_lz/linuxhost_lz/"
User="root"
Password="{AES256:keyfile:a}vHBiEqmmvylNPs90Dw1AhQ==" />
<ConfigFile
Server="CENTRAL-CONFIG-SERVER"
Name="AAGP.xml"
Path="/"
Disp="AAGP" />
</ConfigurationArtifact>
- Save the load list in Install_dir/localconfig/lz/lz_cnfglist.xml.
This directory is the default location on Linux systems and can be
changed with the IRA_SERVICE_INTERFACE_CONFIG_LOADLIST environment
variable.
- Start the Linux OS agent.
All of the central configuration clients that connect
to this central configuration server can now use the
credentials for "itmuser" rather than "root". <ConfigServer
Name="CENTRAL-CONFIG-SERVER"
URL="http://linuxhost:1920///linuxhost_lz/linuxhost_lz/"
User="itmuser"
Password="{AES256:keyfile:a}vHBiEqmmvylNPs90Dw1AhQ==" />
- Cascading central configuration servers
- Other monitoring agents can collect an Access Authorization Group Profile (AAGP) update from a central configuration server and then be used to distribute
files to other agents.
- Cascading servers must be configured so that they distribute configuration load lists that specify their Agent Service Interface for the ConfigServer element's
URL.
- Cascading servers use the DISP=CUSTOM ConfigFile elements to download
the content that they distribute to other agents.