WIP.
Basic tool to generate big-ass LDAP directories.
I was unsatisfied with what I found in the wild to accomplish this task.
Specifically, it seems that everyone has been focused on large user counts but not necessarily complex structures like deep OUs, many users in many groups, things like that.
So this is my attempt to address those needs.
Outputs to stdout. Pipe, redirect, do whatever you need to.
$ ./ldifgen generate --help
NAME:
ldifgen generate - generate ldif file
USAGE:
ldifgen generate [command options] [arguments...]
OPTIONS:
--users value number of users to generate (default: 10)
--ous value number of organizational units to generate (default: 2)
--ou-depth value depth of generated OUs. specify n>1 to create 'chains' of OUs (default: 1)
--groups value number of groups to generate (default: 2)
--domain value domain used to generate DC components, e.g. dc=domain,dc=example,dc=org (default: "domain.example.org")
--user-classes value comma-separated list of classes for user objects (default: "top,person,organizationalPerson,inetOrgPerson")
--ou-classes value comma-separated list of classes for organizational unit objects (default: "top,organizationalUnit")
--group-classes value comma-separated list of classes for group objects (default: "top,groupOfNames")
--user-change-type value LDIF changetype for users (default: "add")
--group-change-type value LDIF changetype for groups (default: "add")
--ou-change-type value LDIF changetype for OUs (default: "add")
--buzzword-dataset value path to an alternative list of buzzwords, used in group generation. provide list of words, separated by newlines
--department-dataset value path to an alternative list of department names, used in OU generation. provide list of words, separated by newlines
--first-name-dataset value path to an alternative list of first names, used in user generation. provide list of words, separated by newlines
--last-name-dataset value path to an alternative list of last names, used in user generation. provide list of words, separated by newlines
--groups-dataset value path to an alternative list of group names, used in group generation. provide list of words, separated by newlines
--help, -h show help (default: false)
- Random name generation for all objects. Pre-seeded with templates as seen in
static/directory, but you can supply your own via the--xxx-datasetarguments. - Prevents name collisions via retry logic. Very basic, just cycles until hitting 5 collisions, then bails.
- Generates simple OU trees, which if seen graphically are really just a single linked list currently. Room for improvement there.
- Generates groups and adds even weighting of users to each group. Future goal: user-definable weighting.
- Customizable change types per object
- Customizable object classes for OUs, groups, users
It can build a 100k-user, 1000-group, 20-OU (at a depth of 20 sub-OUs) in 10 seconds.
$ time ./ldifgen generate --users 100000 --groups 1000 --ous 20 --ou-depth 20 > out.ldif
INFO[0000] OU name collision: Finance. Generating new OU
INFO[0000] OU name collision: Arts and Sciences. Generating new OU
INFO[0000] OU name collision: Center for Computational Science and Engineering. Generating new OU
INFO[0000] OU name collision: Anthropology. Generating new OU
INFO[0000] group name collision. generating new group
real 0m10.994s
user 0m4.061s
sys 0m7.278s
- Allow for creation of custom object types?
- Allow for weighting of users across groups (currently just weights evenly)
- Allow for broader OU trees, currently they're the equivalent of a linked list
- Custom templates?