Skip to content

Latest commit

 

History

History
171 lines (115 loc) · 6.06 KB

File metadata and controls

171 lines (115 loc) · 6.06 KB

ANNOUNCE: On 2020-02-10 the ansible-keylime-tpm-emulator repo has been renamed to keylime-vagrant-ansible-tpm-emulator . If you have a fork you might want to rename the fork just to keep your sanity (although it's not required). You might also consider updating your git remotes, although Github redirect for a while

Keylime Vagrant Ansible TPM Emulator

Build Status Slack chat

A Vagrant file to easily bring up a test Keylime environment using an Ansible role to deploy Keylime with a pre-configured and ready to use TPM Emulator.

For details on using Keylime, please consult the general project documentation

Security Warning

Do not use a software TPM emulator in a production environment.

SELinux is set to permissive for this role.

This role is designed to enable development environment provisioning or to set up a sandbox environment to test drive Keylime.

Should you want to deploy with a hardware TPM, use the anisble-keylime role

Usage: Ansible role

The Ansible role may be used on its own.

Run the example playbook against your target remote node(s). For instance:

ansible-playbook -i your_hosts playbook.yml

Usage: Vagrant

A Vagrantfile is available for provisioning virtual machines for local testing.

Clone the repository and then simply run with the following additional args added to the vagrant command:

  • --instances: The number of Keylime virtual machines to create. If not provided, it defaults to 1
  • --repo: This is intended to help you hack on Keylime. It mounts a local Keylime Git repository into the virtual machine, allowing you to test your code within the VM. This is optional and will mount the repo directory you pass in at "/root/keylime-dev".
  • --cpus: The number of CPUs. If not provided, defaults to 2
  • --memory: The amount of memory to assign. If not provided, defaults to 2048
  • --qualityoflife: Adds a few extras, such as the Powerline improved bash shell prompt as well as an ls alias (ll for ls -lAh). This is optional.

Deployment example, using libvirt as the virtualization provider:

vagrant --instances=2 --repo=/home/jdoe/keylime --cpus=4 --memory=4096  up --provider libvirt --provision

Deployment example, using VirtualBox as the virtualization provider:

vagrant --instances=2 --repo=/home/jdoe/keylime --cpus=4 --memory=4096  up --provider virtualbox --provision
NOTE: Customized args (--instances, --repos etc), come before the main Vagrant args (such as up, status, --provider). Example: To ssh into the second machine instance, keylime2, use the Vagrant command as such : vagrant --instances=2 ssh keylime2

If you would like to customise these defaults without having to specify them on the command line each time, you can use a vagrant_variables.yml file. The simplest way to do this is to copy vagrant_variables.yml.sample to vagrant_variables.yml and edit it:

cp vagrant_variables.yml.sample vagrant_variables.yml

You can still override the defaults in vagrant_variables.yml by using the command line options.

Once the VM is started, use vagrant ssh to ssh into the VM and run sudo su - to become root.

The TPM emulator will be running.

You can then start the various components using commands:

keylime_verifier

keylime_registrar

keylime_agent

keylime_tenant

Note: you will most likely need to export the right TPM2TOOLS_TCTI environment variable before being able to successfully start keylime_agent. To do so: export TPM2TOOLS_TCTI="mssim:port=2321"

A full, step-by-step guide for provisioning a Keylime development setup on Fedora can be found here.

Upgrading VMs

If you just want to upgrade Keylime within your VM(s), running the following as root, from within /root/keylime, should be enough: git pull python setup.py install

To fully rebuild your VM(s), run the following from the directory where you cloned this repo: vagrant destroy Note: this will delete your Keylime VM(s).

You can then re-deploy the VM(s) by re-running the provisioning step.

Lastly, if you have a VM that was provisioned using an older version of Fedora (say, 31, while the current Vagrantfile will use Fedora 33), you will need to remove the Fedora 31 cloudbase image before vagrant up --provision will upgrade you to the new version of Fedora, eg: vagrant box remove fedora/31-cloud-base

WebApp

The web application can be started with the command keylime_webapp. If using Vagrant, port 443 will be forwarded from the guest to port 8443 on the host.

This will result in the web application being available at the following URL:

https://localhost:8443/webapp/

IMA Policy

This role deploys a basic ima-policy into /etc/ima/ima-policy so that IMA run time integrity may be used. For this to activate, you must reboot the machine first (if you're using vagrant, perform vagrant reload)

If for some reason the TPM and IMA emulation aren't working correctly, you can restart those services with

systemctl restart tpm_emulator
systemctl restart ima_emulator

Access to Keylime components from the host

To allow direct access to the Keylime components from the host machine, you can forward the ports for the various Keylime components by uncommenting the relevant lines in the Vagrantfile.

License

Apache 2.0

Contribute

We welcome contributions and pull requests are welcome!

Please ensure CI tests pass!

Contributors