Smart Data Transaction for CPS Test Documentation

Introduction

This document describes the blueprint test environment for the Smart Data Transaction for CPS blueprint. The test results and logs are posted in the Akraino Nexus at the link below:

https://nexus.akraino.org/content/sites/logs/fujitsu/job/sdt/r6/

Akarino Test Group Information

N/A

Testing has been carried out at Fujitsu Limited labs without any Akraino Test Working Group resources.

Overall Test Architecture

Tests are carried out on the architecture  shown in the diagram below.

Test Bed

The test bed consists of a VM running on x86 hardware, performing jump host/deploy and master node roles, two edge nodes on ARM64 (Jetson Nano) hardware, and two sensor nodes on ARM32 (Raspberry Pi) hardware.

A second VM is used to run the BluVal test framework components outside the system under test.

Test Framework

BluVal and additional tests are carried out using Robot Framework.

Traffic Generator

N/A

Test API description

CI/CD Regression Tests: Docker Private Registry

This set of test cases confirms the Docker private registry setup, population, and tear-down procedures.

The Test inputs

The test scripts and data are stored in the source repository's cicd/tests/docker directory.

Test Procedure

The test bed is placed in a state with the deploy and master node setup complete, but with Kubernetes, EdgeX, and the private registry not running.

Execute the test scripts:

robot cicd/tests/docker

Expected output

The test script will start the registry, pull upstream images and populate the registry, clean images left over from the pull process, stop the registry, and remove the registry. The robot command should report success for all test cases.

Test Results

Nexus URL: https://nexus.akraino.org/content/sites/logs/fujitsu/job/sdt/r6/lfedge-docker/1/

Pass (5/5 test cases)

CI/CD Regression Tests: Node Setup

This set of test cases confirms the scripting to initialize master and edge nodes.

The Test inputs

The test scripts and data are stored in the source repository's cicd/tests/install directory.

Test Procedure

The test bed is place in a state where only the deploy node is initialized. No EdgeX or Kubernetes services are running. For a complete test, the master and edge nodes should not have any software installed that was not installed as part of the OS installation.

Execute the test scripts:

robot cicd/tests/install

Expected output

The test scripts will initialize the master and edge nodes and verify the required software is installed. The robot command should report success for all test cases.

Test Results

Nexus URL: https://nexus.akraino.org/content/sites/logs/fujitsu/job/sdt/r6/lfedge-install/1/

Pass (2/2 test cases)

CI/CD Regression Tests: Cluster Setup & Teardown

These test cases verify that the Kubernetes cluster can be initialized, edge nodes added to it and removed, and the cluster torn down.

The Test inputs

The test scripts and data are stored in the source repository's cicd/tests/cluster directory.

Test Procedure

The test bed is placed in a state where all nodes are prepared with required software and the Docker registry is running. The registry must be populated with the Kubernetes and Flannel images from upstream.

Execute the test scripts:

robot cicd/tests/cluster

Expected output

The test scripts will start the cluster, add all configured edge nodes, remove the edge nodes, and reset the cluster. The robot command should report success for all test cases.

Test Results

Nexus URL: https://nexus.akraino.org/content/sites/logs/fujitsu/job/sdt/r6/lfedge-cluster/7/

Pass (4/4 test cases)

CI/CD Regression Tests: EdgeX Services

These test cases verify that the EdgeX micro-services can be started and that MQTT messages are passed to the master node from the services.

The Test inputs

The test scripts and data are stored in the source repository's cicd/tests/edgex directory.

Test Procedure

The test bed is placed in a state where the cluster is initialized and all edge nodes have joined. The Docker registry and mosquitto MQTT broker must be running on the master node. The registry must be populated with all upstream images and custom images. Either the device-lora service should be enabled with dht2lra service running on the sensor nodes, or device-virtual should be enabled to provide readings.

Execute the test scripts:

robot cicd/tests/edgex

Expected output

The test scripts will start the EdgeX micro-services on all edge nodes, confirm that MQTT messages are being delivered from the edge nodes, and stop the EdgeX micro-services. The robot command should report success for all test cases.

Test Results

Nexus URL: https://nexus.akraino.org/content/sites/logs/fujitsu/job/sdt/r6/edgex-install/7/

Pass (8/8 test cases)

CI/CD Regression Tests: LoRa Device Service

These test cases verify that the LoRa device service can read sensor data over the LoRa communications channel.

The Test inputs

The test steps and data are contained in the scripts in the source repository cicd/tests/lora directory.

Test Procedure

The test bed is initialized to the point of having all EdgeX services running, with device-lora enabled.

The dht2lra service is started on the two sensor nodes.

Execute the test scripts:

robot cicd/tests/lora

Expected output

The test cases will check if MQTT messages containing temperature data gathered from the sensor nodes are arriving at the master node on the topic for each each edge node, validating that the LoRa device support is functioning.

The Robot Framework should report success for all test cases.

Test Results

Nexus URL: https://nexus.akraino.org/content/sites/logs/fujitsu/job/sdt/r6/edgex-lora/3/

Pass (2/2 test cases)

Feature Project Tests

N/A

BluVal Tests

BluVal tests for Lynis, Vuls, and Kube-Hunter were executed on the test bed.

The Test inputs

Bluval User Guide

Steps To Implement Security Scan Requirements

https://vuls.io/docs/en/tutorial-docker.html

Test Procedure

1. Deploy a Test VM
2. Copy the folder ~/.kube from Kubernetes master node to the Test VM
3. Create SSH Key to access Kubernetes master node

Vuls

We use Ubuntu 20.04, so we ran Vuls test as follows:

1. Create directory

   $ mkdir ~/vuls $ cd ~/vuls $ mkdir go-cve-dictionary-log goval-dictionary-log gost-log

2. Fetch NVD

   $ docker run --rm -it \ -v $PWD:/go-cve-dictionary \ -v $PWD/go-cve-dictionary-log:/var/log/go-cve-dictionary \ vuls/go-cve-dictionary fetch nvd

3. Fetch OVAL

   $ docker run --rm -it \ -v $PWD:/goval-dictionary \ -v $PWD/goval-dictionary-log:/var/log/goval-dictionary \ vuls/goval-dictionary fetch ubuntu 16 17 18 19 20

4. Fetch gost

   $ docker run --rm -i \ -v $PWD:/gost \ -v $PWD/gost-log:/var/log/gost \ vuls/gost fetch ubuntu

5. Create config.toml

   [servers] [servers.master] host = "192.168.51.22" port = "22" user = "test-user" keyPath = "/root/.ssh/id_rsa" # path to ssh private key in docker

6. Start vuls container to run tests

   $ docker run --rm -it \ -v ~/.ssh:/root/.ssh:ro \ -v $PWD:/vuls \ -v $PWD/vuls-log:/var/log/vuls \ -v /etc/localtime:/etc/localtime:ro \ -v /etc/timezone:/etc/timezone:ro \ vuls/vuls scan \ -config=./config.toml

7. Get the report

   $ docker run --rm -it \ -v ~/.ssh:/root/.ssh:ro \ -v $PWD:/vuls \ -v $PWD/vuls-log:/var/log/vuls \ -v /etc/localtime:/etc/localtime:ro \ vuls/vuls report \ -format-list \ -config=./config.toml

Lynis/Kube-Hunter

1. Create ~/validation/bluval/bluval-sdtfc.yaml to customize the Test

   blueprint: name: sdtfc layers: - os - k8s os: &os - name: lynis what: lynis optional: "False" k8s: &k8s - name: kube-hunter what: kube-hunter optional: "False"

2. Update ~/validation/bluval/volumes.yaml file

   volumes: # location of the ssh key to access the cluster ssh_key_dir: local: '/home/ubuntu/.ssh' target: '/root/.ssh' # location of the k8s access files (config file, certificates, keys) kube_config_dir: local: '/home/ubuntu/kube' target: '/root/.kube/' # location of the customized variables.yaml custom_variables_file: local: '/home/ubuntu/validation/tests/variables.yaml' target: '/opt/akraino/validation/tests/variables.yaml' # location of the bluval-<blueprint>.yaml file blueprint_dir: local: '/home/ubuntu/validation/bluval' target: '/opt/akraino/validation/bluval' # location on where to store the results on the local jumpserver results_dir: local: '/home/ubuntu/results' target: '/opt/akraino/results' # location on where to store openrc file openrc: local: '' target: '/root/openrc' # parameters that will be passed to the container at each layer layers: # volumes mounted at all layers; volumes specific for a different layer are below common: - custom_variables_file - blueprint_dir - results_dir hardware: - ssh_key_dir os: - ssh_key_dir networking: - ssh_key_dir docker: - ssh_key_dir k8s: - ssh_key_dir - kube_config_dir k8s_networking: - ssh_key_dir - kube_config_dir openstack: - openrc sds: sdn: vim:

3. Update ~/validation/tests/variables.yaml file

   ### Input variables cluster's master host host: <IP Address> # cluster's master host address username: <username> # login name to connect to cluster password: <password> # login password to connect to cluster ssh_keyfile: /root/.ssh/id_rsa # Identity file for authentication

4. Run Blucon

   $ bash validation/bluval/blucon.sh sdtfc

Expected output

BluVal tests should report success for all test cases.

Test Results

Vuls results (manual) Nexus URL: https://nexus.akraino.org/content/sites/logs/fujitsu/job/sdt/r6/sdt-vuls/2/

Lynis results (manual) Nexus URL: https://nexus.akraino.org/content/sites/logs/fujitsu/job/sdt/r6/sdt-lynis/3/

Kube-Hunter results Nexus URL: https://nexus.akraino.org/content/sites/logs/fujitsu/job/sdt/r6/sdt-bluval/2/

Vuls

Nexus URL: https://nexus.akraino.org/content/sites/logs/fujitsu/job/sdt/r6/sdt-vuls/2/

There are 17 CVEs with a CVSS score >= 9.0. These are exceptions requested here:

Release 6: Akraino CVE and KHV Vulnerability Exception Request

Lynis

Nexus URL (run via Bluval, without fixes): https://nexus.akraino.org/content/sites/logs/fujitsu/job/sdt/r6/sdt-bluval/2/

Nexus URL (manual run, with fixes): https://nexus.akraino.org/content/sites/logs/fujitsu/job/sdt/r6/sdt-lynis/2/

The initial results compare with the Lynis Incubation: PASS/FAIL Criteria, v1.0 as follows.

The Lynis Program Update test MUST pass with no errors.

2022-03-04 15:33:28 Test: Checking for program update...
2022-03-04 15:33:31 Current installed version  : 301
2022-03-04 15:33:31 Latest stable version      : 307
2022-03-04 15:33:31 Minimum required version   : 297
2022-03-04 15:33:31 Result: newer Lynis release available!
2022-03-04 15:33:31 Suggestion: Version of Lynis outdated, consider upgrading to the latest version [test:LYNIS] [details:-] [solution:-]

Fix: Download and run the latest Lynis directly on SUT.

Steps To Implement Security Scan Requirements#InstallandExecute

The following list of tests MUST complete as passing

Results after the above fixes are as follows:

The Lynis Program Update test MUST pass with no errors.

2022-03-07 15:19:07 Test: Checking for program update...
2022-03-07 15:19:10 Current installed version : 308
2022-03-07 15:19:10 Latest stable version : 307
2022-03-07 15:19:10 No Lynis update available.

The following list of tests MUST complete as passing

The post-fix manual logs can be found at https://nexus.akraino.org/content/sites/logs/fujitsu/job/sdt/r6/sdt-lynis/3/.

Kube-Hunter

Nexus URL (with fixes): https://nexus.akraino.org/content/sites/logs/fujitsu/job/sdt/r6/sdt-bluval/2/

There are 5 Vulnerabilities.

• KHV002
• KHV005
• KHV050
• CAP_NET_RAW Enabled
• Access to pod's secrets

Fix for KHV002

$ kubectl replace -f - <<EOF
apiVersion: rbac.authorization.k8s.io/v1
kind: ClusterRole
metadata:
  annotations:
    rbac.authorization.kubernetes.io/autoupdate: "false"
  labels:
    kubernetes.io/bootstrapping: rbac-defaults
  name: system:public-info-viewer
rules:
- nonResourceURLs:
  - /healthz
  - /livez
  - /readyz
  verbs:
  - get
EOF

Fix for KHV005, KHV050, Access to pod's secrets

$ kubectl replace -f - <<EOF
apiVersion: v1
kind: ServiceAccount
metadata:
  name: default
  namespace: default
automountServiceAccountToken: false
EOF

The above fixes are implemented in the Ansible playbook deploy/playbook/init_cluster.yml and configuration file deploy/playbook/k8s/fix.yml

Fix for CAP_NET_RAW Enabled:

Create a PodSecurityPolicy with requiredDropCapabilities: NET_RAW. The policy is shown below. The complete fix is implemented in the Ansible playbook deploy/playbook/init_cluster.yml and configuration files deploy/playbook/k8s/default-psp.yml and deploy/playbook/k8s/system-psp.yml, plus enabling PodSecurityPolicy checking in deploy/playbook/k8s/config.yml.

apiVersion: policy/v1beta1
kind: PodSecurityPolicy
metadata:
  name: psp-baseline
spec:
  privileged: true
  allowPrivilegeEscalation: true
  allowedCapabilities:
  - IPC_LOCK
  - NET_ADMIN
  requiredDropCapabilities:
  - NET_RAW
  hostIPC: true
  hostNetwork: true
  hostPID: true
  hostPorts:
  - max: 65535
    min: 0
  readOnlyRootFilesystem: false
  fsGroup:
    rule: 'RunAsAny'
  runAsUser:
    rule: 'RunAsAny'
  seLinux:
    rule: 'RunAsAny'
  supplementalGroups:
    rule: 'RunAsAny'
  volumes:
  - '*'

Results after fixes are shown below:

Note that in spite of all Kube-Hunter vulnerabilities being fixed, the results still show one test failure. The "Inside-a-Pod Scanning" test case reports failure, apparently because the log ends with "Kube Hunter couldn't find any clusters" instead of "No vulnerabilities were found." Because vulnerabilities were detected and reported earlier by this test case, and those vulnerabilities are no longer reported, we believe this is a false negative, and may be caused by this issue: https://github.com/aquasecurity/kube-hunter/issues/358

Test Dashboards

Single pane view of how the test score looks like for the Blue print.

*Vuls is counted as one test case.

*One Kube-Hunter failure is counted as a pass. See above.

Vuls and Lynis test cases are failing, an exception request is filed for Vuls-detected vulnerabilities that cannot be fixed. The Lynis results have been confirmed to pass the Incubation criteria.

Additional Testing

None at this time.

Bottlenecks/Errata

None at this time.