ci: add workflow to check for vulnerabilities in images

.github/workflows/sec-scan.yml

@@ -0,0 +1,135 @@
+---
+# The aim of this GitHub workflow is to update the `ci/securitty-scan/security_scan_results.md` with latest security scan results.
+name: Update notebook image security reports
+on:
+  workflow_dispatch:
+    inputs:
+      branch:
+        required: true
+        description: "Provide the name of the branch you want to update ex main, vYYYYx etc: "
+        # Put the scheduler on comment until automate the full release procedure
+        # schedule:
+        #   - cron: "0 0 * * 5"  #Scheduled every Friday
+env:
+  SEC_SCAN_BRANCH: sec-scan-${{ github.run_id }}
+  BRANCH_NAME: ${{ github.event.inputs.branch || 'main' }}
+  RELEASE_VERSION_N: 2023b
+  RELEASE_VERSION_N_1: 2023a
+jobs:
+  initialize:
+    runs-on: ubuntu-latest
+    permissions:
+      contents: write
+    steps:
+      - name: Install Skopeo CLI
+        shell: bash
+        run: |
+          sudo apt-get -y update
+          sudo apt-get -y install skopeo
+
+      # Checkout the branch
+      - name: Checkout branch
+        uses: actions/checkout@v3
+        with:
+          ref: ${{ env.BRANCH_NAME }}
+
+      # Create a new branch
+      - name: Create a new branch
+        run: |
+         echo ${{ env.SEC_SCAN_BRANCH }}
+         git checkout -b ${{ env.SEC_SCAN_BRANCH }}
+         git push --set-upstream origin ${{ env.SEC_SCAN_BRANCH }}
+
+  update-n-version:
+    needs: [initialize]
+    runs-on: ubuntu-latest
+    permissions:
+      contents: write
+    steps:
+      - name: Configure Git
+        run: |
+         git config --global user.email "github-actions[bot]@users.noreply.github.com"
+         git config --global user.name "GitHub Actions"
+
+      # Get the latest weekly build commit hash: https://github.com/opendatahub-io/notebooks/commits/2023b
+      - name: Checkout upstream notebooks repo
+        uses: actions/checkout@v3
+        with:
+          repository: opendatahub-io/notebooks.git
+          ref: ${{ env.RELEASE_VERSION_N }}
+
+      - name: Retrieve latest weekly commit hash from the release branch
+        id: hash-n
+        shell: bash
+        run: |
+          echo "HASH_N=$(git rev-parse --short HEAD)" >> ${GITHUB_OUTPUT}
+
+      - name: Checkout "N - 1" branch
+        uses: actions/checkout@v3
+        with:
+          repository: opendatahub-io/notebooks.git
+          ref: ${{ env.RELEASE_VERSION_N_1 }}
+
+      - name: Retrieve latest weekly commit hash from the release branch
+        id: hash-n-1
+        shell: bash
+        run: |
+          echo "HASH_N_1=$(git rev-parse --short HEAD)" >> ${GITHUB_OUTPUT}
+
+      # Checkout the release branch to apply the updates
+      - name: Checkout release branch
+        uses: actions/checkout@v3
+        with:
+          ref: ${{ env.SEC_SCAN_BRANCH }}
+
+      - name: setup python
+        uses: actions/setup-python@v4
+        with:
+          python-version: '3.10' # install the python version needed
+
+      - name: install python packages
+        run: |
+          python -m pip install --upgrade pip
+          pip install requests
+
+      - name: execute py script # run trial.py
+        env:
+          HASH_N : ${{ steps.hash-n.outputs.HASH_N }}
+          RELEASE_VERSION_N: 2023b
+
+          HASH_N_1: ${{ steps.hash-n-1.outputs.HASH_N_1 }}
+          RELEASE_VERSION_N_1: 2023a
+        run: make scan-image-vulnerabilities 
+
+      - name: Push the files
+        run: |
+              git fetch origin  ${{ env.SEC_SCAN_BRANCH }} && git pull origin  ${{ env.SEC_SCAN_BRANCH }} && git add . && git commit -m "Update security scans" && git push origin  ${{ env.SEC_SCAN_BRANCH }}
+
+  # Creates the Pull Request
+  open-pull-request:
+    needs: [update-n-version]
+    runs-on: ubuntu-latest
+    permissions:
+      pull-requests: write
+    steps:
+      - name: Checkout repo
+        uses: actions/checkout@v3
+
+      - name: pull-request
+        uses: repo-sync/pull-request@v2
+        with:
+          source_branch: ${{ env.SEC_SCAN_BRANCH }}
+          destination_branch: ${{ env.BRANCH_NAME}}
+          github_token: ${{ secrets.GH_TOKEN }}
+          pr_label: "automated pr"
+          pr_title: "[Security Scanner Action] Weekly update of security vulnerabilities reported by Quay"
+          pr_body: |
+            :rocket: This is an automated Pull Request.
+
+            This PR updates:
+
+            * `ci/security-scan/security_scan_results.md` file with the latest security vulnerabilities reported by Quay.
+            * `ci/security-scan/weekly_commit_ids` with the latest updated SHA digests of the notebooks (N & N-1)
+            Created by `/.github/workflows/sec-scan.yaml`
+
+            :exclamation: **IMPORTANT NOTE**: Remember to delete the ` ${{ env.SEC_SCAN_BRANCH }}` branch after merging the changes

Makefile

@@ -468,4 +468,9 @@ refresh-pipfilelock-files:
 	cd runtimes/tensorflow/ubi8-python-3.8 && pipenv lock
 	cd runtimes/tensorflow/ubi9-python-3.9 && pipenv lock
 	cd base/c9s-python-3.9 && pipenv lock
-
+
+# This is only for the workflow action
+# For running manually, set the environment variables "RELEASE_VERSION_N"  and "HASH_N"
+.PHONY: scan-image-vulnerabilities
+scan-image-vulnerabilities:
+	python ci/security-scan/quay_security_analysis.py

ci/security-scan/quay_security_analysis.py

@@ -0,0 +1,162 @@
+import os
+import subprocess
+import re
+from datetime import date
+import requests
+from collections import Counter
+import fileinput
+
+my_dictionary = {}
+
+commit_id_path = "ci/security-scan/weekly_commit_ids.env"
+
+
+IMAGES = [
+    "odh-minimal-notebook-image-n",
+    "odh-runtime-minimal-notebook-image-n",
+    "odh-runtime-data-science-notebook-image-n",
+    "odh-minimal-gpu-notebook-image-n",
+    "odh-pytorch-gpu-notebook-image-n",
+    "odh-generic-data-science-notebook-image-n",
+    "odh-tensorflow-gpu-notebook-image-n",
+    "odh-trustyai-notebook-image-n",
+    "odh-habana-notebook-image-n",
+    "odh-codeserver-notebook-n",
+    "odh-rstudio-notebook-n",
+    "odh-rstudio-gpu-notebook-n"
+]
+
+IMAGES_N_1 = [
+    "odh-minimal-notebook-image-n-1",
+    "odh-minimal-gpu-notebook-image-n-1",
+    "odh-pytorch-gpu-notebook-image-n-1",
+    "odh-runtime-data-science-notebook-image-n-1",
+    "odh-generic-data-science-notebook-image-n-1",
+    "odh-tensorflow-gpu-notebook-image-n-1",
+    "odh-trustyai-notebook-image-n-1",
+    "odh-habana-notebook-image-n-1",
+    "odh-codeserver-notebook-n-1",
+    "odh-rstudio-notebook-n-1",
+    "odh-rstudio-gpu-notebook-n-1"
+]
+
+def process_image(image, commit_id_path, RELEASE_VERSION_N, HASH_N):
+    with open(commit_id_path, 'r') as params_file:
+        img_line = next(line for line in params_file if re.search(f"{image}=", line))
+        img = img_line.split('=')[1].strip()
+
+    registry = img.split('@')[0]
+
+    src_tag_cmd = f'skopeo inspect docker://{img} | jq \'.Env[] | select(startswith("OPENSHIFT_BUILD_NAME=")) | split("=")[1]\''
+    src_tag = subprocess.check_output(src_tag_cmd, shell=True, text=True).strip().strip('"').replace('-amd64', '')
+
+
+    regex = "" # f"{src_tag}-{RELEASE_VERSION_N}-\\d+-{HASH_N}"
+
+    if RELEASE_VERSION_N == "":
+        regex = f"{src_tag}-(\\d+-)?{HASH_N}"
+    else:
+        regex = f"{src_tag}-{RELEASE_VERSION_N}-\\d+-{HASH_N}"
+
+    latest_tag_cmd = f'skopeo inspect docker://{img} | jq -r --arg regex "{regex}" \'.RepoTags | map(select(. | test($regex))) | .[0]\''
+    print("latest_tag_cmd")
+    print(latest_tag_cmd)
+    latest_tag = subprocess.check_output(latest_tag_cmd, shell=True, text=True).strip()
+
+    digest_cmd = f'skopeo inspect docker://{registry}:{latest_tag} | jq .Digest | tr -d \'"\''
+    digest = subprocess.check_output(digest_cmd, shell=True, text=True).strip()
+
+    output = f"{registry}@{digest}"
+
+    print("output")
+    print(output)
+
+    sha_ = output.split(":")[1]
+
+    url = f"https://quay.io/api/v1/repository/opendatahub/workbench-images/manifest/sha256:{sha_}/security"
+    headers = {
+        "X-Requested-With": "XMLHttpRequest",
+        "Authorization": "Bearer 3PZX0UYX6FSENKQ14I1VTHUJ4KGBS8L5LHJ0W1RN7TPHFVQ4P0NR7VQNCZIFRC9B_1"
+    }
+
+    response = requests.get(url, headers=headers)
+    data = response.json()
+
+    vulnerabilities = []
+
+    for feature in data['data']['Layer']['Features']:
+        if(len(feature['Vulnerabilities']) > 0):
+            for vulnerability in feature['Vulnerabilities']:
+                vulnerabilities.append(vulnerability)
+
+    severity_levels = [entry.get("Severity", "Unknown") for entry in vulnerabilities]
+    severity_counts = Counter(severity_levels)
+
+    my_dictionary[latest_tag] = {}
+    my_dictionary[latest_tag]['sha']= digest
+
+    for severity, count in severity_counts.items():
+        my_dictionary[latest_tag][severity] = count
+
+    for line in fileinput.input(commit_id_path, inplace=True):
+        if line.startswith(f"{image}="):
+            line = f"{image}={output}\n"
+        print(line, end="")
+
+
+RELEASE_VERSION_N = os.environ['RELEASE_VERSION_N']
+HASH_N = os.environ['HASH_N']
+
+# Call the function for each image in IMAGES
+for i, image in enumerate(IMAGES):
+    process_image(image, commit_id_path, RELEASE_VERSION_N, HASH_N)
+
+today = date.today()
+d2 = today.strftime("%B %d, %Y")
+
+formatted_data = ""
+for key, value in my_dictionary.items():
+    formatted_data += f"| [{key}](https://quay.io/repository/opendatahub/workbench-images/manifest/{my_dictionary[key]['sha']}?tab=vulnerabilities) |"
+    for severity in ['Medium', 'Low', 'Unknown', 'High', 'Critical']:
+        count = value.get(severity, 0)  # Get count for the severity, default to 0 if not present
+        formatted_data += f" {count} |"
+    formatted_data += "\n"
+
+my_dictionary = {}
+
+RELEASE_VERSION_N_1 = os.environ['RELEASE_VERSION_N_1']
+HASH_N_1 = os.environ['HASH_N_1']
+
+for i, image in enumerate(IMAGES_N_1):
+    process_image(image, commit_id_path, RELEASE_VERSION_N_1, HASH_N_1)
+
+branch_n_data = ""
+for key, value in my_dictionary.items():
+    branch_n_data += f"| [{key}](https://quay.io/repository/opendatahub/workbench-images/manifest/{my_dictionary[key]['sha']}?tab=vulnerabilities) |"
+    for severity in ['Medium', 'Low', 'Unknown', 'High', 'Critical']:
+        count = value.get(severity, 0)  # Get count for the severity, default to 0 if not present
+        branch_n_data += f" {count} |"
+    branch_n_data += "\n"
+
+markdown_content = """# Security Scan Results
+
+Date: {todays_date}
+
+# Branch N
+
+| Image Name | Medium | Low | Unknown | High | Critical |
+|------------|-------|-----|---------|------|------|
+{table_content}
+
+# Branch N - 1
+
+| Image Name | Medium | Low | Unknown | High | Critical |
+|------------|-------|-----|---------|------|------|
+{branch_n}
+"""
+
+final_markdown = markdown_content.format(table_content=formatted_data, todays_date=d2, branch_n=branch_n_data)
+
+# Writing to the markdown file
+with open("ci/security-scan/security_scan_results.md", "w") as markdown_file:
+    markdown_file.write(final_markdown)