This is the QGG stateless ldap server.
docker run -d -p 389:389 \
-e CONFIG_REPO=bitbucket.org/hpchud/ldap-config.git \
-e CONFIG_USER=hpchud
-e CONFIG_PASS=<api key> \
hpchud/ldap
The data that is used to populate the ldap server must be in another repository and passed to the container through environment variable CONFIG_REPO. This repo must have the following layout:
./cn=config/
olcDatabase\=\{2\}hdb.ldif
olcDatabase\=\{1\}monitor.ldif
<etc>
./populate/
users.ldif
groups.ldif
<etc>
On startup, this repository will be cloned using CONFIG_USER and CONFIG_PASS as the credentials. Whilst this image can be public, you don't want the configuration to be.
The cn=config/*.ldif files will be applied to configure the server, against the first database (0).
The populate/*.ldif files will be applied to populate the server, against the next available database.
When the server is restarted, it is discarded completely - a brand new server is configured, populated again with the users and groups.
Thus, any changes performed against the server whilst it is running will be discarded. All changes should be made to the ldif file in the prepopulate folder.
The benefits are many; the most important one being that no complex replication policies are required for high availability - just start up more than 1 instance, configure the client appropriately, and remember to restart them one by one when updating.