A set of Ansible playbooks to configure a freshly installed Cisco Identity Services Engine (ISE) for simple operation; in my case, a basic Cisco Software-Defined Access environment.
UPDATE: I have redone this project using Python in case you want to compare methods. That project is available here.
These playbooks will configure the following in ISE:
- local user groups (
01_add_groups.yaml) - local user identities (
02_add_users.yaml) - a simple TACACS profile and command set for privilege 15 access (
03_create_tacacs_profiles.yaml) - TACACS policies in the default policy set (
05_create_tacacs_authz_policies.yaml) - Scalable Group Tags (SGT) to allow our authentication rules to work (
06_create_sgts.yaml) - network access authorization rules to places users in the appropriate VLANs (wired and wireless) (
08_create_authorization_profiles.yaml) - network access policies to authorize users and assign SGTs (
09_create_authorization_policies.yaml) - a complete wired guest workflow with redirection, portal, and SGT(
10_create_guest_authz_profiles.yaml&11_create_guest_authz_policies.yaml) - Cisco access point profiling (using the wired guest flow) and authorization profiles (
12_access_point_profiling.yaml)
The resources that are configured with these playbooks are enough to support a basic Cisco SD-Access network including:
- TACACS authentication for network devices
- dot1x authentication and authorization for multiple users
- wired guest access
- multiple Scalable Group Tags (SGTs)
- Cisco access point profiling and authorization
I administer a lab environment that is used to demonstrate Cisco Software-Defined Access for customers. When new versions of Cisco ISE or DNA Center are released, I do a fresh installation of both so that I can test the new versions with the lab workflow. This involves installing each piece of software and then configuring them both to the point where I can start going through the lab guide.
After watching a demo of the collections in this repo that use Terraform and Ansible to spin-up and configure ISE in AWS, I was inspired to setup something similar to assist in my configuration process when testing new versions.
I started with almost zero API experience beyond installing Postman on my workstation in the past and never using it. Prior to this project I had run exactly one Ansible playbook in my life, and that was six years ago. Needless to say, I was (and still am) completely green with this stuff, so it was a complete learning experience for me, especially not having a background in code or data structures. To that end, I have tried to structure and comment the playbooks to make it easy for someone like me to pick up and get working without much effort.
As a bonus: You will notice some snark in the playbook comments as well, which stemmed from some frustrations that I ran into while learning.
- Cisco Identity Services Engine (ISE) 3.1 or higher
Note: Some of these playbooks may work with ISE 3.0, but 3.1 is required for the policy stuff.
- Python 3.6+
- Ansible 2.9 or higher
- Cisco ISE SDK v1.0.0+
- Ansible Modules for Cisco ISE
- JMESPath for JSON filtering
See the instructions on the Ansible Modules for Cisco ISE site for more installation details.
If you just want to see these in action, you can run them against a Cisco DevNet ISE 3.1 APIs, Ansible, and Automation sandbox instance without any customization:
- On your local workstation install requirements above:
Ansible:
sudo pip install ansible
Cisco ISE SDK:
sudo pip install ciscoisesdk
Ansible ISE Collection:
ansible-galaxy collection install cisco.ise
JMESPath:
pip install jmespath
-
Reserve a sandbox in DevNet and connect to it per their instructions
-
In ISE, enable ERS and Open API settings in: Administration | Settings | API Settings | API Service Settings
- Run the playbooks one at a time like this:
$ ansible-playbook -i hosts 01_add_groups.yaml
$ ansible-playbook -i hosts 02_add_users.yaml
$ ansible-playbook -i hosts 03_create_tacacs_profiles.yaml
- You can verify the changes in the ISE GUI after each playbook if you're curious
Although my use-case for these playbooks involves a fresh deployment of ISE to support a Cisco SD-Access topology, they can absolutely be modified and used in a brownfield ISE environment without SDA.
I'm going to try to make the project self-documenting via comments as best I can, but here's a rough guide to get started:
credentials.yaml - Contains the ISE deployment information such as hostname, username, and password
groupsandusers.yaml - Contains the internal identity groups and users that will be configured by the playbooks
policy.yaml - Contains the policy/profile information that will be configured by the playbooks
-
I developed these playbooks using Cisco DevNet's ISE 3.1 APIs, Ansible, and Automation sandbox, so they should be useable there without any modification (See Quick Start section)
-
If you're using macOS you may receive this error when running your playbook:
objc[34120]: +[__NSCFConstantString initialize] may have been in progress in another thread when fork() was called.
objc[34120]: +[__NSCFConstantString initialize] may have been in progress in another thread when fork() was called. We cannot safely call it or ignore it in the fork() child process. Crashing instead. Set a breakpoint on objc_initializeAfterForkError to debug.
ERROR! A worker was found in a dead state
Try setting these environment variables:
export OBJC_DISABLE_INITIALIZE_FORK_SAFETY=YES
export no_proxy=*
One day I will post a summary of some of the ISE settings that I change to make my life a little easier following an install. These settings will be pretty specific to a lab environment and not suggested for production.
better documentation- even more better documentation
- really good documentation
- documents
- better comments in the playbooks
- better optimization of the playbooks
separate environment-specific values into dedicated files for easier customization- add more optional fields to make this useful in the real world
redo this whole mess in Python before I retire(NOTE: I beat the deadline - the Python version of these playbooks is available here)
This was definitely a learning exercise for me and just something I wanted to see if I could pull off on my own. That said, I want to thank John Capobianco for helping point me in the right direction for certain things and only pressuring me to use Python twelve times a day instead of the usual twenty. I also want to thank Josh Bronikowski for helping me figure out that the ISE API requires extra fields for some calls, and Ryan Malone for some JSON tips.
Please open an issue if you have any questions or suggestions.
I developed these playbooks for my own use, so I do want to keep them as clean as I can, but if you think they can be improved or optimized, feel free to submit a PR.