New Setup Design


Introduction

SetupUtil was originally created to:

Some of this work was done by directory server instance creation. The code in the old cfg_sspt.c and configure_instance.cpp was a mix of directory server specific code and code needed by setuputil to create the initial o=NetscapeRoot tree and database.

Setuputil also provides a (woefully incomplete) C++ wrapper class around the LDAP C SDK which somewhat simplifies LDAP interactions. It also provides native charset <-> UTF8 conversion.

Plan

Remove setuputil. Replace its existing functionality.

DS base package

Add perl modules for setup. These modules will handle the command line interactive UI phase. This will allow us to have an interactive UI for setting up base directory server instances, and act as a wrapper around ds_newinst.

DS admin package

The DS admin package will extend the base perl modules to add support for configuration DS instance creation and configuration, for registering instances with the config DS, and for admin server configuration.

o=NetscapeRoot entry creation

This section shows how to make this happen: “Almost all of the o=NetscapeRoot entry creation can be done with LDIF templates with key values replaced at set up time by perl scripts.”

Perl script register_server.pl processes the tasks. The script uses the Setup Inf module as well as PerLDAP.

    Usage: $bindir/register_server.pl [ -Fv ] [ -h `<host>` ] [ -p `<port>` ] [ -D `<rootdn>` ]`
                              -w `<rootdnpw>` [ -d `<default_infdir>` ]`
                              -i `<inffile(s)>` ... -m `<mapfile>` `<ldiffile>` ...`
    Description: Store server info stored in the ldiffiles to the Configuration`
                 Directory Server replacing the macros with the defined values`
                 in the map file.`
    -H: help (print this message)`
    -v: verbose`
    -F: Fresh registration; i.e., removing the existing server info.`
    -h `<host>`: configuration server host (localhost, by default)`
    -p `<port>`: configuration server port (389)`
    -D `<rootdn>`: configuration server's rootdn ("cn=Directory Manager")`
    -w `<rootdnpw>`: configuration server's rootdn password`
    -d `<default_infdir>`: the directory where static .inf files are located ("$infdir")`
    -i `<inffile(s)>`: dynamic .inf file(s)`
    -m `<mapfile>`: map file name`
    <ldiffile>` ...: ldif files or template ldif files to be stored in the Configuration Directory Server`

map file: $infdir/register_param.map needs to be specified, which defines the key - value pairs replaced at the run time. This file is a .in file in the adminserver source tree. The @…@ parameters are processed at the build time.

    This file is used by the register_server.pl script to register the server
    info to the Configuration Directory Server.  The server info is stored in
    the (template) ldif files located in @ldifdir@. In case a server entry has
    %...% format parameters, this map table is used to resolve it and replace
    the parameter with the value defined in this file.
    [Parameter resolution rules]
    * If the right-hand value is in ` (backquote), the value is eval'ed by perl. 
      The output should be stored in $returnvalue to pass to the internal hash.
    * If the right-hand value is in " (doublequote), the value is passed as is.
    * If the right-hand value is not in any quote, the value should be found
      in either of the setup inf file (static) or the install inf file (dynamic).
    * Variables surrounded by @ (e.g., /export/servers/ds72/etc/fedora-ds/admin-serv) are replaced with the
      system path at the compile time.
    * The right-hand value can contain variables surrounded by % (e.g., %asid%)
      which refers the right-hand value (key) of this map file.

ldif template files: $ldifdir/[0-9][0-9]filename.ldif.tmpl The ldif tmpl files contain ldif formatted entries including %…% style parameters. All the parameters must be defined in the map file. sample entry in a ldif tmpl file:

    dn: ou=%domain%, o=NetscapeRoot
    objectClass: top
    objectClass: organizationalUnit
    objectClass: nsadmindomain
    ou: %domain%
    description: Standard branch for configuration information
    nsAdminDomainName: %domain%

In addition, two kinds of .inf files are needed to fill in the parameters. static .inf files: installed at $infdir in the adminserver installation dynamic .inf file(s): temporary file created by the Setup module dynamically; one file per Directory Server

Sample usages:

    Clean Configuration Directory Server; register one server
    register_server.pl -m infdir/register_param.map -w rootdnpasswd -i /dir/for/dynamic/inf/setup###0.inf ldifdir/*.ldif.tmpl


    Register 2 more servers' info to the Configuration Directory Server
    register_server.pl -m infdir/register_param.map -w rootdnpasswd -i "/dir/for/dynamic/inf/setup###1.inf \
    /dir/for/dynamic/inf/setup###2.inf" ldifdir/*.ldif.tmpl

    Register 3 servers' info to the Configuration Directory Server overriding the existing o=netscaperoot if any
    register_server.pl -F -m infdir/register_param.map -w rootdnpasswd -i "/dir/for/dynamic/inf/setup###0.inf \
    /dir/for/dynamic/inf/setup###1.inf /dir/for/dynamic/inf/setup###2.inf" ldifdir/*.ldif.tmpl
Last modified on 29 November 2024