How to Write a lib389 Test


Lib389 is a python base library for managing Directory servers. It was initially created to help writing tests of DS. It can also used to create new administration CLIs. This document is focusing on writing tests for the Directory Server. Lib389 can be used to generate individual testcases(tickets), and suites that test a range of functionality/feature. When working on a new bug/ticket that is reproducible, a testcase should be written that can reproduce the problem. Then that same testcase can be used to verify the fix. The entire collection of testcases(tickets) can then be used to check for regressions in future releases/builds.

When a new feature is added, a testing suite should be created. This suite is in its own dedicated directory, and can contain multiple scripts, or whatever is required to effectively test a feature.

Directory Server’s Testing Layout

After you checkout the source code, you can find the testing framework as follows:

mkdir /home/<USER>/ds-source
cd /home/<USER>/ds-source
git clone
cd ds/dirsrvtests

Under ds/dirsrvtests/ we see the following directories

suites/  --> location for functional/feature tests

tickets/ --> location for individual ticket/bug tests

data/    --> location to store static files needed by tickets/suites.
             This is accessed by calling: "tmp_dir = topology.standalone.getDir(__file__, TMP_DIR)"

tmp/     --> location where the lib389 tests can use to write temporary files to
             This is accessed by calling: "data_dir = topology.standalone.getDir(__file__, DATA_DIR)"

Get the Lib389 Framework

To start working with lib389, we need to get the lib389 framework:

mkdir /home/<USER>/lib389
cd /home/<USER>/lib389
git clone

External users will use “git clone git://” instead of “git clone ssh://

Next, we add the location of the framework to our PYTHONPATH environment variable (/home/mreynolds/lib389):


Deploying the Directory Server

There are two options for testing the Directory Server:

RPM Installation

To run the lib389 tests against an RPM installation, you must be logged in as ‘root’.

Private Build of the Directory Server

There are actually two options here as well. You can build, and install the server into the RPM location.

Or you can build and install the Directory Server into a “prefixed/specific” location. In this case you would set PREFIX to the Directory Server installation directory:


For more information on how to make/install the Directory Server into a prefixed/specific location see:

Upstream Setup Script


Directory Server Build Page

There are also helper scripts listed on the upstream testing reference page:

Create a Test

To write a lib389 test script we to mention the default layout of a ticket. Each test needs these basic functions, but there could be more depending on your needs:

Example using a single standalone instance:

class TopologyStandalone(object):
    def __init__(self, standalone):
        self.standalone = standalone

def topology(request):
    # Topology function
    # Create our instances, configure replication

    return TopologyStandalone(standalone)

def test_ticket#####_run(topology):
    # Test function
    # You can create as many of these mfunctions as you like

def test_ticket#####_final(topology):
    # Final function
    topology.standalone.delete()'Testcase PASSED')

def run_isolated():
    # Run function
    topo = topology(True)

if __name__ == '__main__':

Fortunately, we have a helper script( to create this template for us, so we can focus on writing the actual test

Using this script we can setup complex replication instances(that are fully configured and initialized), or multiple standalone instances. The script is located in the Directory Server source code:


./ -t|--ticket <ticket number> | -s|--suite <suite name> [ i|--instances <number of standalone instances> [ -m|--masters <number of masters> -h|--hubs <number of hubs> -c|--consumers <number of consumers> ] -o|--outputfile ]

Currently the script will create a “replication” setup, or a standalone setup (no replication). When using the “-o|–outputfile” option be sure to use _Here is an example of both:

./ --ticket 33333

    --> Creates a script named:
        This script create a single standalone instance

./ --ticket 33333 --outputfile

    --> Creates a script named:
        This script create a single standalone instance

./ --ticket 33333 --instances 4

    --> Creates a script named:
        This script creates four standalone instances

./ --ticket 33333 --master 2

    --> Creates a script named:
        This script creates two replication master instances, and initializes them

./ --ticket 33333 --master 2 --hubs 2 --consumers 10

    --> Creates a script named:
        This script creates two replication master instances, two replication hubs, and ten consumers.  Replication is fully initialized across all the servers

./ --suite memberOf_plugin
    --> Creates a "suite" test named:

Note - “suite” tests are for testing functional areas or features, while “ticket” tests are for testing individual bugs.

If setting up a complex replication test, the script structures the deployment in a cascading fashion:

             master1  <=>  master2
                   hub1  hub2

     consumer  consumer  consumer consumer

The Defaults

lib389 uses many predefined variables to ease test writing. These are just a few, the complete list is in “” in the lib389 framework

Common Operation Examples:

Here are some examples of common operations that are performed in a test script

Creating/Deleting Instances

Creating instances takes a few steps listed below

Start, Stop, and Restart the Server


Add, Modify, and Delete Operations

# Add an entry
USER_DN = 'cn=mreynolds,%s' % DEFAULT_SUFFIX
    topology.standalone.add_s(Entry((USER_DN, {
                              'objectclass': 'top person'.split(),
                              'cn': 'mreynolds',
                              'sn': 'reynolds',
                              'userpassword': 'password'
except ldap.LDAPError, e:
    log.error('Failed to add user (%s): error (%s)' % (USER_DN, e.message['desc']))
    assert False

# Modify an entry
    topology.standalone.modify_s(USER_DN, [(ldap.MOD_REPLACE, 'cn', 'Mark Reynolds')])
except ldap.LDAPError, e:
    log.error('Failed to modify user (%s): error (%s)' % (USER_DN, e.message['desc']))
    assert False

# Delete an entry
except ldap.LDAPError, e:
    log.error('Failed to delete user (%s): error (%s)' % (USER_DN, e.message['desc']))
    assert False

Search and Bind Operations

By default when an instance is create and opened(, it is already authenticated as the Root DN(Directory Manager). So you can just start searching without having to “bind”

# Search - By default we are already authenticated as "cn=directory manager".
#          We are also only requesting the 'cn' attribute
    entries = topology.standalone.search_s(DEFAULT_SUFFIX, ldap.SCOPE_SUBTREE, '(cn=*)', ['cn'])
    for entry in entries:
        entry.hasValue('cn', 'Mark Reynolds'):
  'Search found "Mark"')
except ldap.LDAPError, e:
    log.fatal('Search failed, error: ' + e.message['desc'])
    assert False

# Bind - bind as our test entry
    topology.standalone.simple_bind_s(USER_DN, "password")
except ldap.LDAPError, e:
    log.error('Bind failed for (%s), error (%s)' % (USER_DN, e.message['desc']))
    assert False

# Rebind as the Root DN
    topology.standalone.simple_bind_s(DN_DM, PASSWORD)
except ldap.LDAPError, e:
    log.error('Bind failed for (%s), error (%s)' % (DN_DM, e.message['desc']))
    assert False

Plugin Enabling/Disabling

Enable/Disable a plugin. See for all the plugin name variables(like PLUGIN_MEMBER_OF)


Adding Tasks

There are some built task functions, that add the task and wait for it to finish. See for all the available Slapi Tasks.

args{TASK_WAIT: True} tells the function to wait for the task to complete

    topology.standalone.tasks.fixupMemberOf(suffix=SUFFIX, args={TASK_WAIT: True})
except ValueError:
    log.error('Some problem occured with a value that was provided')
    assert False

Here is another way to run a task

importTask = Tasks(topology.standalone)
args = {TASK_WAIT: True}
    importTask.importLDIF(DEFAULT_SUFFIX, None, ldif_file, args)
except ValueError:
    assert False

Configuring Replication

After the instance is created, you can enable it for replication and set up a replication agreement.

Memory Leak Testing (valgrind)

These are the core module functions that make valgrind testing very simple.


You must stop all the server instances before runing valgrind_enable(). Here is an example:

# First stop the instance

# Get the sbin directory so we know where to replace 'ns-slapd'
sbin_dir = get_sbin_dir(prefix=topology.standalone.prefix)

# Enable valgrind

Then you must start the instance(s) before running your tests

valgrind_check_leak(dirsrv_inst, pattern)

You must provide the DirSrv instance object so we know what process to check as there could be multiple Directory Server processes running, and then the text or regular expression you want to look for in the valgrind output. This function stops the instance you want to check, as that’s the only way to get the output file written to disk. Here is an example:

if valgrind_check_leak(topology.standalone, 'range_candidates'):
    log.error('test_range_search: Memory leak is still present!')

# topology.standalone is now stopped


This restores the original ns-slapd binary to the sbin directory. It’s important to note that when doing valgrind tests you should always call valrgind_disable() before exiting the script or calling assert False, as it will leave the valgrind wrapper in the sbin directory and it could break other tests.

Here is an example:

# Disable valgrind
sbin_dir = get_sbin_dir(prefix=topology.standalone.prefix)

Running Tests

There are two ways to runs the test, run a single test, or test with “py.test” which can run all the tests it can find, etc. See “py.test –help” for more details.

Running a Single Test

If running the test on an RPM, you need to be root. It will deploy the instance under ‘/’ and run them as user/group ‘dirsrv’. If testing against a specific installation(non-RPM) you need to set the PREFIX to its location:

Running as a local user against a specific installation location

PREFIX=/home/mreynolds/install python ./

Running as ‘root’ against the RPM installation

python ./

Running All The Tests

We use “py.test” to run multiple tests. “py.test” will select all the files/testcases containing a ‘test’ string. Here are few examples

# This runs all the tests in ds/dirsrvtests/tickets/
cd ds/dirsrvtests/tickets
PREFIX=/home/mreynolds/install py.test -v

# This runs all the tests for debugging: -x -s
PREFIX=/home/mreynolds/install py.test -v -x -s

# Run the 'suites' and 'tickets' tests
cd ds/dirsrvtests/
PREFIX=/home/mreynolds/install py.test -v

For debugging the tests use “-x” and “-s

py.test -v -x -s

-x Stops the run when an error occurs
-s Writes the script output to stdout, instead of suppressing it

The Developer Expectation

Run all the tests before committing a patch. Write your own test - the more tests the better. Let’s build quality software with no regressions!


Run “pydoc lib389” to see the entire modules documentation

Also checkout the Upstream Testing Framework page:


Last modified on 21 March 2017