Using bcfg2-info

bcfg2-info is a tool for introspecting server functions. It is useful for understanding how the server is interpreting your repository. It consists of the same logic executed by the server to process the repository and produce configuration specifications, just without all of the network communication code. Think of bcfg2-info as bcfg2-server on a stick. It is a useful location to do testing and staging of new configuration rules, prior to deployment. This is particularly useful when developing templates, or developing Bcfg2 plugins.

Getting Started

First, fire up the bcfg2-info interpreter.

[0:464] bcfg2-info
Loading experimental plugin(s): Packages
NOTE: Interfaces subject to change
Handled 8 events in 0.006s
Handled 4 events in 0.035s
Welcome to bcfg2-info
Type "help" for more information
>

At this point, the server core has been loaded up, all plugins have been loaded, and the bcfg2-info has both read the initial state of the Bcfg2 repository, as well as begun monitoring it for changes. Like bcfg2-server, bcfg2-info monitors the repository for changes, however, unlike bcfg2-server, it does not process change events automatically. File modification events can be processed by explicitly calling the update command. This will process the events, displaying the number of events processed and the amount of time taken by this processing. If no events are available, no message will be displayed. For example, after a change to a file in the repository:

> update
Handled 1 events in 0.001s
> update
>

This explicit update process allows you to control the update process, as well as see the precise changes caused by repository modifications.

bcfg2-info has several builtin commands that display the state of various internal server core state. These are most useful for examining the state of client metadata, either for a single client, or for clients overall.

clients

Displays a list of clients, along with their profile groups

groups

Displays a list of groups, the inheritance hierarchy, profile status, and category name, if there is one.

showclient

Displays full metadata information for a client, including profile group, group memberships, bundle list, and any connector data, like Probe values or Property info.

config

Displays the configuration of the Bcfg2 server.

To leave the interactive shell, just type quit or exit.

Debugging Configuration Rules

In addition to the commands listed above for viewing client metadata, there are also commands which can shed light on the configuration generation process. Recall that configuration generation occurs in three major steps:

1. Resolve client metadata
2. Build list of entries for the configuration
3. Bind host-specific version of each entry

Step 1 can be viewed with the commands presented in the previous section. The latter two steps can be examined using the following commands.

showentries

displays a list of entries (optionally filtered by type) that appear in a client’s configuration specification

buildbundle

Render a single bundle template. This only performs the template rendering step; it does not fully bind all entries in the bundle. This command is very useful when developing bundle templates.

buildfile

Perform the entry binding process on a single entry, displaying its results. This command is very useful when developing configuration file templates.

build

Build the full configuration specification and write it to a file.

mappings

displays the entries handled by the plugins loaded by the server core. This command is useful when the server reports a bind failure for an entry.

Debugging and Developing Bcfg2

bcfg2-info loads a full Bcfg2 server core, so it provides the ideal environment for developing and debugging Bcfg2. Because it is hard to automate this sort of process, we have only implemented two commands in bcfg2-info to aid in the process.

profile

The profile command produces python profiling information for other bcfg2-info commands. This can be used to track performance problems in configuration generation.

debug

The debug command exits the bcfg2-info interpreter loop and drops to a python interpreter prompt. The Bcfg2 server core is available in this namespace as “self”. Full documentation for the server core is out of scope for this document. This capability is most useful to call into plugin methods, often with setup calls or the enabling of diagnostics.

It is possible to return to the bcfg2-info command loop by exiting the python interpreter with ^D.

There is built-in support for IPython in bcfg2-info. If IPython is installed, dropping into debug mode in bcfg2-info will use the IPython interpreter by default.