Commit a3598893 authored by Stephen Morris's avatar Stephen Morris
Browse files

[3398] Miscellaneous corrections and edits as part of review

parent 83b2d2b5
......@@ -14,99 +14,99 @@
/**
@page configBackend Kea configuration backends
@page configBackend Kea Configuration Backends
@section configBackendIntro Configuration backend introdution
@section configBackendIntro Introduction
Kea is a flexible DHCP protocol engine. It offers a selection of lease database
backends, extensibility via hooks API and definition of custom options.
backends, extensibility via the hooks API and the definition of custom options.
Depending on the environment, one lease database backend may be better than
other. Similarly, depending on the deployment, it would also make sense to
provide different ways to configure the server. The capability to have different
ways to configure the servers is called Configuration Backend. As the way
how configuration is received cannot be part of the configuration itself, it
has to be chosen at the compilation time (when configuring sources). This page
explains why we chose that path and how it is implemented. It can be used by
people who want to develop and maintain their own configuration backends.
@section configBackendMotivation Motivation for configuration backends
BIND10 used to maintain an extensive framework that was responsible for
components configuration. After BIND10 was cancelled, two projects were
created: Kea and Bundy. Kea team decided to remove BIND10 framework, while Bundy
team decided to keep it. Even though Kea team is focused on JSON backend, which
reads the JSON configuration file from disk, we try to make it easy for others
to use different backends.
other. Similarly, because the best way of configuring the server can depend on
the environment, Kea provides different ways of obtaining configuration
information, through the Configuration Backend. Since the means by which
configuration information is received cannot be part of the configuration itself, it
has to be chosen at the compilation time (when configuring the sources).
This page explains the background to the Configuration Backend and how
it is implemented. It is aimed at people who want to develop and
maintain their own backends.
@section configBackendMotivation Motivation for Different Backends
BIND10 (the project under which the first stages of Keas were developed)
used to maintain an extensive framework that was responsible for the
configuration of components. After BIND10 was cancelled, two projects
were created: <a href="http://kea.isc.org">Kea</a> (focused on DHCP)
and <a href="http://www.bundy-dns.de">Bundy</a> (aimed at DNS). The
Kea team decided to remove the BIND10 framework, while the Bundy team
decided to keep it. However, even though the Kea team is focused on a
backend that reads a JSON configuration file from disk, it decided to
make it easy for others to use different backends.
While ISC currently (May 2014) plans to maintain only one configuration backend
(JSON: a JSON file read from disk), there may be other organizations (e.g.
possibly Bundy project community) that will maintain other backends. It is quite
possible that other configuration backends (e.g. using LDAP or XML) will be
(a JSON file read from disk), there may be other organizations (e.g.
the Bundy project community) that will maintain other backends. It is quite
possible that additional backends (e.g. using LDAP or XML) will be
developed and maintained by other organizations.
@section configBackendAdding How to add a new configuration backend?
@section configBackendAdding How to Add a New Configuration Backend
@todo Will be covered in ticket #3400.
@todo: Will be covered in ticket #3400.
@section configBackendJSONDesign The JSON Configuration Backend
@section configBackendJSONDesign Design for JSON configuration backend
The following are some considerations that shaped the design of the configuration
backend framework.
-# A new parameter called --with-kea-config will be implemented in
configure script. It will allow selecting at compilation time how the
-# A new parameter called --with-kea-config will be implemented in the
configure script. It will allow the selection at compilation time of how the
servers will be configured. For the next 2-3 months (until around June 2014),
we'll have 2 values: JSON (read from file) and BIND10 (use bind10 framework).
Once we have file based configuration implemented and we're ready to switch
there will be two values: JSON (read from file) and BIND10 (use the BIND10 framework).
Once the file based configuration is implemented and the Kea team is ready to switch
(i.e. enough confidence, Forge tests updated for new configuration
mechanism), BIND10 backend will be removed from Kea repo. Other projects
(e.g. Bundy) who want to maintain it, are advisaged to just revert a single
commit that will bring back the BIND10 framework to their repos.
This switchable backend concept is really simple. There are just different
implementations of ControlledXSrv class, so it's a matter with compiling/linking
one file or another. Hence it is easy for us to remove the old backend (and for
Bundy folks to keep it if they desire so). It is also easy for other
organizations to add and maintain their own backends (e.g. LDAP based).
For detailed description of DHCPv6 backend, see @ref dhcpv6ConfigBackend.
-# Retain config and command callbacks. Each backend must use the common code
mechanism), the BIND10 backend will be removed from the Kea repository. Other projects
(e.g. Bundy) who want to maintain it, are advised to just revert the single
commit that will bring the BIND10 framework back to their repositories.<br/><br/>
This switchable backend concept is quite simple. There are just different
implementations of ControlledXSrv class, so it is a matter of compiling/linking
one file or another. Hence it is easy to remove the old backend (and for
Bundy to keep it if they desire so). It is also easy for other
organizations to add and maintain their own backends (e.g. LDAP).<br/><br/>
-# Each backend must use the common code
for configuration and command processing callbacks. They all assume that
JSON formatted parameters are sent and they are expected to return well
formatted JSON responses. Exact format of configuration and commands is
module specific.
-# After Kea 0.9 is released, we will design some form of secure socket that
we'll be able to send commands over. Whatever the design we end up with, it
will allow to send configs and commands in JSON format and get responses.
Once that is done, we'll have the same capability as we did in BIND10
framework: to send additional parameters. One obvious use case will be
to send new config file name as parameter for "reload".
-# We need to add command handler for reading config from a file. Its main
responsibility is to load config from file and process it. The JSON backend
must call that handler when starting up the server.
-# Extend existing JSON parser. We need to extend current JSON parser in
@ref isc::data::Element::fromJSON() to allow optional preprocessing.
For now that capability will simply remove hash comments, but it is expected
formatted JSON responses. The exact format of configuration and commands is
module specific.<br/><br/>
-# After Kea 0.9 is released, a form of secure socket will be implemented through
which commands can be sent. Whatever the design, it
will allow the sending of configurations and commands in JSON format and
the receiving of responses.<br/><br/>
Once that is done, Kea will have the same capability the BIND10
framework to send additional parameters. One obvious use case will be
to send a new configuration file name as the parameter for "reload".<br/><br/>
-# A command handler needs to be added for reading the configuration from a file. Its main
responsibility is to load the configuration and process it. The JSON backend
must call that handler when starting up the server.<br/><br/>
-# Extend the existing JSON parser. The current JSON parser in
@ref isc::data::Element::fromJSON() needs to be extended to allow optional preprocessing.
For now that capability will simply remove whole-line comments staring with the hash
character, but it is expected
to grow over time (in-line comments and file inclusions are the obvious
envisaged additions).
-# Implement common base class for Kea4, Kea6, D2 server. Some operations will be
common for all 3 components: logger initialization, handling, and some time
later control socket. This calls for a small base class that Dhcpv4Srv,
Dhcpv6Srv and D2Controller can use. We will start that base class (@ref
isc::dhcp::Daemon) as very small one. It is expected to grow over time as we
do more code unification.
-# We need to implement a way to initialized stand-alone logging (i.e. each
Kea component will initialize it on its own).
-# Config file format.
We will use the current format of b10-config.db. This is slight change
to what we did in Kea during BIND10 days, because we were receiving a subset
of that configuration. Let me give specific example. That's how b10-config.db
looks like today:
envisaged additions).<br/><br/>
-# Implement a common base class for the Kea4, Kea6, and D2 servers. Some operations will be
common for all three components: logger initialization, handling and, at some future point,
control socket. This calls for a small base class that @ref isc::dhcp::Dhcpv4Srv "Dhcpv4Srv",
@ref isc::dhcp::Dhcpv6Srv "Dhcpv6Srv" and the @ref isc::d2::D2Controller "D2Controller" classes can use.
It is expected that the base class
(@ref isc::dhcp::Daemon) will be a small one but will grow over time as the code is unified.<br/><br/>
-# A way is needed to initialize stand-alone logging (i.e. each
Kea component will initialize it on its own).<br/><br/>
-# The current format of the BIND10 configuration file, b10-config.db will be
retained as the configuration file format. This is slight change
from the BIND10 days, as then a subset of the configuration was received by
the daemon processes.<br/><br/>
To take a specific example, the following is how b10-config.db
looks today:<br/><br/>
@code
{
"Init": { ... }
......@@ -127,38 +127,30 @@ developed and maintained by other organizations.
}
}
@endcode
Kea components used to receive only relevant parts of it (e.g. Kea4
received config that contained content of the Dhcp4 element). We'll be
receiving the whole config now. The modification in the code is really
minor: just iterate over top level elements and pick the appropriate
tree (or get element by name). Also, that approach makes the logging
initialization code very easy to share among Kea4, Kea6 and D2.
-# We keep .spec files. We'll keep and maintain them even though we won't do
anything with them. Those files were used by bindctl to do syntax checking.
We will be lacking that capability for a while. Implementing C++ code for
.spec validation of received config is out of scope for 0.9 (and probably
for 1.0 as this is pretty big task).
-# Shell script to start/stop Kea4,Kea6 and D2. There will be a script that will
start, stop and reconfigure the daemons. It will be rather simple. Its only
job will be to pass config file to each daemon and remember its PID file, so
sending signals would be possible (for config reload or shutdown). Optionally,
<br/>
The Kea components used to receive only relevant parts of it (e.g. Kea4
received config that contained content of the Dhcp4 element). Now they
will receive all of it. The modification in the code to handle this
is really minor: just iterate over the top level elements and pick the appropriate
tree (or get the element by name). Also, that approach makes the logging
initialization code very easy to share among Kea4, Kea6 and D2.<br/><br/>
-# The .spec files used in BIND 10 by the control program to validate commands
will be retained. They will be kept and maintained even though no use of
them is planned. At some future time syntax validation may be implemented,
although it is out of scope for Kea 0.9 (and probably
for 1.0 as it is pretty big task).<br/><br/>
-# Addition of a shell script to start/stop Kea4,Kea6 and D2. There will be a script that will
start, stop and reconfigure the daemons. Its only
job will be to pass the configuration file to each daemon and remember its PID file, so
that sending signals will be be possible (for configuration reload or shutdown). Optionally,
it could also print out a status based on PID, but that may be tricky to
implement in a portable way. The minimum set of commands would be:
implement in a portable way. The minimum set of commands will be:
-# Start the processes
- eventually based on config, initially start them all
- eventually based on configuration, initially start them all
- it could launch a nanny script which restarts them on a crash (past 0.9)
-# Prompt the processes to reload configuration
- for now it will be a matter of sending singal to the right process
- this could also decide if D2 should still be running or not,
and react accordingly (past 0.9)
- this could also decide if D2 should still be running or not, and react accordingly (post 0.9)
-# Stop the processes in an orderly fashion
-# Perhaps return status of each process
*/
\ No newline at end of file
*/
Markdown is supported
0% or .
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment