4 | Chapter 4 | |
Configuring Distributed Pollers |
Distributed pollers are configured by the administrator on the master station. A poller is a computer in the network which can talk to local devices and report the result to the master station. A distributed poller can be any machine capable of communicating with the local nodes via an interface.
When configuring Distributed Pollers, a strong understanding of how pollers and poller packages are built will help reduce configuration time and eliminate errors.
Poller packages are ways to organize configuration information into reusable containers of information. A poller package contains four basic items:
defines the boundaries of the discovery domain. Devices not listed in a range will be ignored by the distributed poller.
defines the types of devices to discover. Filters are rules which are evaluated TRUE or FALSE when a device is interrogated. If the rule evaluates TRUE, the device will be added to the database of the distributed poller. If the rule evaluates FALSE, the device will be ignored. Filters are combined with ranges to define the boundaries of the management domain and the interesting devices within that domain.
defines the types of services to probe on a device. Services include icmp, http, smtp, or any currently supported protocols.
defines the hours of operation for the package; both one-time and repeating calendars. Periodic backup, preventative maintenance and other scheduled outages are flagged here. (This functionality is curently disabled).
A Distributed Poller can have one or more poller package assigned to it. When a poller package is assigned to a poller, the poller uses the information in the package to control the behavior, parameters and methods of the poller. The package is downloaded to the poller at initialization time. This architecture allows several kinds of configurations:
If a poller package is assigned to more than one poller (currently not supported), the domains defined in that poller package overlap. Should one of the pollers fail, the other poller will continue to operate. Overlapping pollers will generate multiple traffic since both pollers operate against the same set of devices.
If a poller package is assigned to only one poller, then that poller is solely responsible for polling services. Traffic is minimized in this configuration, however, a single poller has a single point of failure.
The administrator builds new pollers, assigns poller packages and (in the near future) sets calendars of operation. To launch the configuration panel for the distributed pollers:
Log in as the administrator. The Administrator Main Panel appears:
![]() |
Figure: Administrator Main Panel |
Select the "Configure Distributed Pollers" icon from the Administrator Main Panel. The
Configure Distributed Pollers panel appears:
![]() |
Figure: Configure Distributed Pollers Panel |
The "Configure Distributed Pollers" panel is used to build or modify distributed pollers, poller packages, and assign packages to pollers. The panel has a menu bar at the top, icon bar for tools, configuration areas and a status bar.
The Configure Distributed Pollers panel is divided into 3 separate areas:
Defines the packages configured on the system. Each package has four branches below it representing the ranges, filters, services and calendars assigned to that package. Packages are dragged onto pollers using drag and drop operations.
The services area contains all of the supported services which a poller could potentially probe. Since a service requires poller code specific to that service, additional services can only be added to the list by adding new service modules to Bluebird.
Pollers are shown in this area of the panel. If a poller is not listed here, it is not polling. Pollers must have at least one poller package assigned to them. Packages are dragged from the Poller Package area and dropped onto a Distributed Poller.
The icon bar holds the more commonly used tools for configuring pollers. Icons exist for copy, paste, expand/contract, wizards and adding new pollers.
![]() |
Figure: Distributed Poller Icon Bar |
- Save the current configuration. Allows the changed configuration
to be make permanent. As changes are made, they are stored in a temporary location until a save if performed.
Exiting the panel without saving changes will generate a confirmation message warning about the loss of
data.
- Copy the selected item to the copy buffer.
A paste is necessary to insert the copied information.
- Paste the last copied information into the selected
item. Pasting one type of information into a different type of information is
not allowed and will be ignored or generate an error message.
- Expand all the folder icons to view the
contents.
- Expand the selected folder icon and show
all the contents. Double clicking on the folder icon will accomplish the same task.
- Collapse all folder icons. All
contents of folder tabs will be hidden.
- Collapse the selected folder icon.
Double clicking on an open folder icon will accomplish the same task.
- Add a new poller icon. This function
is disabled until distributed management capability is enabled.
- Add a new poller package. Selecting
this icon will pop up a panel allowing the poller information to be configured.
- Add a one-time calendar. This
function is disabled until calendar functionality is available.This
- Add a repeating calendar. This
function is disabled until calendar functionality is available.
- Launch the Wizard. Currently disabled
until distributed management capability is available.
The menu bar allows menu operation of basic functions such as copying and pasting. In addition to mouse clicks, keyboard accelerators allow keyboard operation of basic functions. The various menu bar options are shown below:
The file menu allows the current configuration to be saved and the application
to be exited. If modification to the distributed poller configuration has
not been saved, a warning message will appear allowing the configuration to
be saved or discarded.
![]() |
Figure: File Menu Bar |
The Edit menu allows basic delete,copy and paste functions. The selected item in the node
tree is either deleted, copied into the copy storage area or pasted from the copy
storage area.
![]() |
Figure: Edit Menu Bar |
View allows the look and feel of the panel to be chosen. Personal attributes
of the panel are saved on a user by user basis, including look and feel.
![]() |
Figure: View Menu Bar |
The Help menu bar provides basic "About" functionality and entry point
into the help subsystem.
![]() |
Figure: Help Menu Bar |
The right mouse click is used extensively for changing the properies of the various configurations in the poller packages and pollers. Right click functionality is sensitive:
Right clicking on the Ranges branch in the Poller Packages section allows
ranges to be copied, pasted and properties changed.
![]() |
Figure: Right click on "Ranges" |
Right clicking on the Filters branch in the Poller Packages section allows
filters to be copied, pasted and properties changed. Under Properties,
the graphical rule builder is used to define the rules for that filter.
![]() |
Figure: Right click on "Filters" |
Right clicking on the Services branch in the Poller Packages section allows
services to be copied, pasted and properties changed.
![]() |
Figure: Right click on "Services" |
Ranges defined the "responsibility domain" of a poller package. This domain is bounded by the start and stop IP addresses of devices within it's responsibility. If a device is not defined within a range, Bluebird will not manage it. Ranges are additive and subtractive; i.e. you can define multiple ranges and all the ranges are added together to form a superset. In addition, ranges can be excluded to never manage devices within a range.
Careful definition of ranges can allow multiple distributed pollers to overlap responsibility. Packages should be defined to be reusable by different pollers.
When the "Properties" option is selected from a right click on a ranges
folder, the following panel appears:
![]() |
Figure: Ranges Default Panel |
The timeout determines how long Bluebird will wait for a device to respond to an ICMP message before considering the device unreachable.
The retries determine the number of times Bluebird will retry a node which has timed out (see Timeout above) and is unreachable. After the "Retries" number of retries, the node is considered unreachable.
When the "Include Ranges" folder tab is selected, the following panel appears:
![]() |
Figure: Include Ranges Panel |
The beginning IP address in the management range.
The ending IP address in the management range.
The timeout determines how long Bluebird will wait for a device to respond to an ICMP message before considering the device unreachable.
The retries determine the number of times Bluebird will retry a node which has timed out (see Timeout above) and is unreachable. After the "retries" number of retries, the node is considered unreachable.
When the "Exclude Ranges" folder tab is selected, the following panel appears:
![]() |
Figure: Exclude Ranges Panel |
The beginning IP address to be excluded from the management range.
The ending IP address to be excluded from the management range.
When the "Specific Nodes" folder tab is selected, the following panel appears:
![]() |
Figure: Specific Node Panel |
The IP address to be specifically included from the management range.
The timeout determines how long Bluebird will wait for the device to respond to an ICMP message before considering the device unreachable.
The retries determine the number of times Bluebird will retry the node which has timed out (see Timeout above) and is unreachable. After the "retries" number of retries, the node is considered unreachable.
When the "URLs" folder tab is selected, the following panel appears:
![]() |
Figure: URLs Panel |
The URL of a file containing a list of nodes to be included in the management domain.
A file browser allowing you to pick the URL.
The timeout determines how long Bluebird will wait for a device to respond to an ICMP message before considering the device unreachable.
The retries determine the number of times Bluebird will retry a node which has timed out (see Timeout above) and is unreachable. After the "retries" number of retries, the node is considered unreachable.