Refactor container naming: base image default, Python explicit

- Rename SandboxBase → SandboxPython in test worker
- Flip header logic: X-Sandbox-Type: python selects Python image
- Default (no header) now uses base image (smaller, no Python)
- Update cleanup script for -python suffix
- Remove per-run CI cleanup (resources persist until PR closes)

Author: ghostwriternr

.github/workflows/pullrequest.yml

@@ -160,16 +160,10 @@ jobs:
           AWS_ACCESS_KEY_ID: ${{ secrets.AWS_ACCESS_KEY_ID }}
           AWS_SECRET_ACCESS_KEY: ${{ secrets.AWS_SECRET_ACCESS_KEY }}
 
-      # Cleanup: Delete test worker and container (only for PR environments)
-      - name: Cleanup test deployment
-        if: always() && github.event_name == 'pull_request'
-        continue-on-error: true
-        run: |
-          cd tests/e2e/test-worker
-          ../../../scripts/cleanup-test-deployment.sh ${{ steps.env-name.outputs.worker_name }}
-        env:
-          CLOUDFLARE_API_TOKEN: ${{ secrets.CLOUDFLARE_API_TOKEN }}
-          CLOUDFLARE_ACCOUNT_ID: ${{ secrets.CLOUDFLARE_ACCOUNT_ID }}
+      # Note: Resources are NOT cleaned up after each run to speed up subsequent CI runs.
+      # Cleanup happens via:
+      # - cleanup.yml: Triggered when PR is closed
+      # - cleanup-stale.yml: Daily cron job for orphaned/stale resources
 
   # Validate changesets don't contain internal packages
   validate-changesets:

scripts/cleanup-test-deployment.sh

@@ -2,11 +2,15 @@
 set -e
 
 # Cleanup Test Deployment Script
-# Deletes a test worker and its associated container with proper ordering and retry logic
+# Deletes a test worker and its associated containers with proper ordering and retry logic
 #
 # Usage: ./cleanup-test-deployment.sh <worker-name>
 # Example: ./cleanup-test-deployment.sh sandbox-e2e-test-worker-pr-123
 #
+# This script handles multiple container variants:
+# - <worker-name>: Base image container (no Python, default)
+# - <worker-name>-python: Python image container
+#
 # Environment variables required:
 # - CLOUDFLARE_API_TOKEN
 # - CLOUDFLARE_ACCOUNT_ID
@@ -21,30 +25,42 @@ fi
 
 echo "=== Starting cleanup for $WORKER_NAME ==="
 
-# Step 1: Get container ID BEFORE deleting worker (critical order!)
-echo "Looking up container ID..."
+# Step 1: Get container IDs BEFORE deleting worker (critical order!)
+echo "Looking up container IDs..."
 
 # Get container list (wrangler outputs JSON by default, no --json flag needed)
 RAW_OUTPUT=$(npx wrangler containers list 2>&1)
 
+CONTAINER_ID=""
+CONTAINER_PYTHON_ID=""
+
 # Check if output looks like JSON (starts with '[')
 if echo "$RAW_OUTPUT" | grep -q '^\['; then
   echo "✓ Got JSON output from wrangler containers list"
 
-  # Parse JSON to find container
+  # Parse JSON to find both containers
   CONTAINER_ID=$(echo "$RAW_OUTPUT" | jq -r ".[] | select(.name==\"$WORKER_NAME\") | .id" 2>/dev/null || echo "")
+  CONTAINER_PYTHON_ID=$(echo "$RAW_OUTPUT" | jq -r ".[] | select(.name==\"$WORKER_NAME-python\") | .id" 2>/dev/null || echo "")
 
   if [ -n "$CONTAINER_ID" ]; then
-    echo "✓ Found container: $CONTAINER_ID"
+    echo "✓ Found base container: $CONTAINER_ID"
   else
-    echo "⚠️  No container found for worker $WORKER_NAME"
+    echo "⚠️  No base container found for $WORKER_NAME"
+  fi
+
+  if [ -n "$CONTAINER_PYTHON_ID" ]; then
+    echo "✓ Found python container: $CONTAINER_PYTHON_ID"
+  else
+    echo "⚠️  No python container found for $WORKER_NAME-python"
+  fi
+
+  if [ -z "$CONTAINER_ID" ] && [ -z "$CONTAINER_PYTHON_ID" ]; then
     echo "Available containers:"
     echo "$RAW_OUTPUT" | jq -r '.[].name' 2>/dev/null || echo "(unable to parse container names)"
   fi
 else
   echo "⚠️  Non-JSON output from wrangler containers list:"
   echo "$RAW_OUTPUT"
-  CONTAINER_ID=""
 fi
 
 # Step 2: Delete worker
@@ -55,23 +71,41 @@ else
   echo "⚠️  Worker deletion failed or already deleted"
 fi
 
-# Step 3: Delete container with retry logic (if we found one)
-if [ -n "$CONTAINER_ID" ]; then
-  echo "Deleting container with retry logic..."
+# Function to delete a container with retry logic
+delete_container() {
+  local container_id=$1
+  local container_name=$2
+
+  if [ -z "$container_id" ]; then
+    return 0
+  fi
+
+  echo "Deleting $container_name container with retry logic..."
   for i in 1 2 3; do
-    if npx wrangler containers delete "$CONTAINER_ID" 2>/dev/null; then
-      echo "✓ Container deleted successfully"
-      break
+    if npx wrangler containers delete "$container_id" 2>/dev/null; then
+      echo "✓ $container_name container deleted successfully"
+      return 0
     else
       if [ $i -lt 3 ]; then
-        echo "⚠️  Container deletion attempt $i/3 failed, retrying in 5s..."
+        echo "⚠️  $container_name container deletion attempt $i/3 failed, retrying in 5s..."
         sleep 5
       else
-        echo "❌ Container deletion failed after 3 attempts"
-        exit 1
+        echo "❌ $container_name container deletion failed after 3 attempts"
+        return 1
       fi
     fi
   done
+}
+
+# Step 3: Delete containers
+CLEANUP_FAILED=false
+
+delete_container "$CONTAINER_ID" "base" || CLEANUP_FAILED=true
+delete_container "$CONTAINER_PYTHON_ID" "python" || CLEANUP_FAILED=true
+
+if [ "$CLEANUP_FAILED" = true ]; then
+  echo "=== Cleanup completed with errors ==="
+  exit 1
 fi
 
 echo "=== Cleanup complete ==="

tests/e2e/code-interpreter-workflow.test.ts

@@ -27,7 +27,7 @@ import { getTestWorkerUrl, WranglerDevRunner } from './helpers/wrangler-runner';
 import {
   createSandboxId,
   createTestHeaders,
-  createBaseImageHeaders,
+  createPythonImageHeaders,
   cleanupSandbox
 } from './helpers/test-fixtures';
 import type { CodeContext, ExecutionResult } from '@repo/shared';
@@ -64,7 +64,7 @@ describe('Code Interpreter Workflow (E2E)', () => {
 
   test('should create and list code contexts', async () => {
     currentSandboxId = createSandboxId();
-    const headers = createTestHeaders(currentSandboxId);
+    const headers = createPythonImageHeaders(currentSandboxId);
 
     // Create Python context
     const pythonCtxResponse = await fetch(
@@ -112,7 +112,7 @@ describe('Code Interpreter Workflow (E2E)', () => {
 
   test('should delete code context', async () => {
     currentSandboxId = createSandboxId();
-    const headers = createTestHeaders(currentSandboxId);
+    const headers = createPythonImageHeaders(currentSandboxId);
 
     // Create context
     const createResponse = await fetch(`${workerUrl}/api/code/context/create`, {
@@ -154,7 +154,7 @@ describe('Code Interpreter Workflow (E2E)', () => {
 
   test('should execute simple Python code', async () => {
     currentSandboxId = createSandboxId();
-    const headers = createTestHeaders(currentSandboxId);
+    const headers = createPythonImageHeaders(currentSandboxId);
 
     // Create Python context
     const ctxResponse = await fetch(`${workerUrl}/api/code/context/create`, {
@@ -185,7 +185,7 @@ describe('Code Interpreter Workflow (E2E)', () => {
 
   test('should maintain Python state across executions', async () => {
     currentSandboxId = createSandboxId();
-    const headers = createTestHeaders(currentSandboxId);
+    const headers = createPythonImageHeaders(currentSandboxId);
 
     // Create context
     const ctxResponse = await fetch(`${workerUrl}/api/code/context/create`, {
@@ -228,7 +228,7 @@ describe('Code Interpreter Workflow (E2E)', () => {
 
   test('should handle Python errors gracefully', async () => {
     currentSandboxId = createSandboxId();
-    const headers = createTestHeaders(currentSandboxId);
+    const headers = createPythonImageHeaders(currentSandboxId);
 
     // Create context
     const ctxResponse = await fetch(`${workerUrl}/api/code/context/create`, {
@@ -267,7 +267,7 @@ describe('Code Interpreter Workflow (E2E)', () => {
 
   test('should execute simple JavaScript code', async () => {
     currentSandboxId = createSandboxId();
-    const headers = createBaseImageHeaders(currentSandboxId);
+    const headers = createTestHeaders(currentSandboxId);
 
     // Create JavaScript context
     const ctxResponse = await fetch(`${workerUrl}/api/code/context/create`, {
@@ -297,7 +297,7 @@ describe('Code Interpreter Workflow (E2E)', () => {
 
   test('should maintain JavaScript state across executions', async () => {
     currentSandboxId = createSandboxId();
-    const headers = createBaseImageHeaders(currentSandboxId);
+    const headers = createTestHeaders(currentSandboxId);
 
     // Create context
     const ctxResponse = await fetch(`${workerUrl}/api/code/context/create`, {
@@ -337,7 +337,7 @@ describe('Code Interpreter Workflow (E2E)', () => {
 
   test('should handle JavaScript errors gracefully', async () => {
     currentSandboxId = createSandboxId();
-    const headers = createBaseImageHeaders(currentSandboxId);
+    const headers = createTestHeaders(currentSandboxId);
 
     // Create context
     const ctxResponse = await fetch(`${workerUrl}/api/code/context/create`, {
@@ -375,7 +375,7 @@ describe('Code Interpreter Workflow (E2E)', () => {
 
   test('should stream Python execution output', async () => {
     currentSandboxId = createSandboxId();
-    const headers = createTestHeaders(currentSandboxId);
+    const headers = createPythonImageHeaders(currentSandboxId);
 
     // Create context
     const ctxResponse = await fetch(`${workerUrl}/api/code/context/create`, {
@@ -457,7 +457,7 @@ for i in range(3):
 
   test('should process data in Python and consume in JavaScript', async () => {
     currentSandboxId = createSandboxId();
-    const headers = createTestHeaders(currentSandboxId);
+    const headers = createPythonImageHeaders(currentSandboxId);
 
     // Create Python context
     const pythonCtxResponse = await fetch(
@@ -528,7 +528,7 @@ console.log('Sum:', sum);
 
   test('should isolate variables between contexts', async () => {
     currentSandboxId = createSandboxId();
-    const headers = createTestHeaders(currentSandboxId);
+    const headers = createPythonImageHeaders(currentSandboxId);
 
     // Create two Python contexts
     const ctx1Response = await fetch(`${workerUrl}/api/code/context/create`, {
@@ -583,7 +583,7 @@ console.log('Sum:', sum);
 
   test('should maintain isolation across many contexts (12+)', async () => {
     currentSandboxId = createSandboxId();
-    const headers = createBaseImageHeaders(currentSandboxId);
+    const headers = createTestHeaders(currentSandboxId);
 
     // Create 12 contexts
     const contexts: CodeContext[] = [];
@@ -653,7 +653,7 @@ console.log('Sum:', sum);
 
   test('should maintain state isolation with concurrent context execution', async () => {
     currentSandboxId = createSandboxId();
-    const headers = createBaseImageHeaders(currentSandboxId);
+    const headers = createTestHeaders(currentSandboxId);
 
     // Create contexts sequentially
     const contexts: CodeContext[] = [];
@@ -734,7 +734,7 @@ console.log('Sum:', sum);
 
   test('should prevent concurrent execution on same context', async () => {
     currentSandboxId = createSandboxId();
-    const headers = createBaseImageHeaders(currentSandboxId);
+    const headers = createTestHeaders(currentSandboxId);
 
     // Create single context
     const ctxResponse = await fetch(`${workerUrl}/api/code/context/create`, {
@@ -896,7 +896,8 @@ console.log('Sum:', sum);
 
   test('should return helpful error when Python unavailable on base image', async () => {
     currentSandboxId = createSandboxId();
-    const headers = createBaseImageHeaders(currentSandboxId);
+    // Use default headers (base image, no Python) to test Python-not-available error
+    const headers = createTestHeaders(currentSandboxId);
 
     // Try to create Python context on base image (no Python installed)
     const response = await fetch(`${workerUrl}/api/code/context/create`, {

tests/e2e/helpers/test-fixtures.ts

@@ -60,21 +60,21 @@ export function createTestHeaders(
 }
 
 /**
- * Create headers for base image sandbox (without Python)
+ * Create headers for Python image sandbox (with Python)
  *
- * Use this for testing the lean image variant that doesn't include Python.
- * The base image is smaller but only supports JavaScript/shell execution.
+ * Use this for testing the full image variant that includes Python.
+ * The Python image is larger but supports Python code execution.
  *
  * @param sandboxId - Which container instance to use
  * @param sessionId - (Optional) Which session within that container
  */
-export function createBaseImageHeaders(
+export function createPythonImageHeaders(
   sandboxId: string,
   sessionId?: string
 ): Record<string, string> {
   return {
     ...createTestHeaders(sandboxId, sessionId),
-    'X-Sandbox-Type': 'base'
+    'X-Sandbox-Type': 'python'
   };
 }
 

tests/e2e/test-worker/Dockerfile

@@ -1,6 +1,6 @@
-# Integration test Dockerfile
-# Uses the -python variant because E2E tests include Python code execution tests
-FROM docker.io/cloudflare/sandbox-test:0.5.6-python
+# Base image Dockerfile (no Python)
+# Used for testing Python-not-available error handling
+FROM docker.io/cloudflare/sandbox-test:0.5.6
 
 # Expose ports used for testing
 EXPOSE 8080

tests/e2e/test-worker/Dockerfile.base

@@ -0,0 +1,6 @@
+# Integration test Dockerfile
+# Uses the -python variant because E2E tests include Python code execution tests
+FROM docker.io/cloudflare/sandbox-test:0.5.6-python
+
+# Expose ports used for testing
+EXPOSE 8080

tests/e2e/test-worker/Dockerfile.python

@@ -5,10 +5,10 @@
  * Supports both default sessions (implicit) and explicit sessions via X-Session-Id header.
  *
  * Two sandbox types are available:
- * - Sandbox: Full image with Python (default)
- * - SandboxBase: Base image without Python (for testing Python-not-available errors)
+ * - Sandbox: Base image without Python (default, lean image)
+ * - SandboxPython: Full image with Python (for code interpreter tests)
  *
- * Use X-Sandbox-Type header to select: 'base' for SandboxBase, anything else for Sandbox
+ * Use X-Sandbox-Type header to select: 'python' for SandboxPython, anything else for Sandbox
  */
 import { Sandbox, getSandbox, proxyToSandbox } from '@cloudflare/sandbox';
 import type {
@@ -25,14 +25,14 @@ import type {
   ErrorResponse
 } from './types';
 
-// Export Sandbox twice - once as Sandbox (python image) and once as SandboxBase (base image)
+// Export Sandbox twice - once as Sandbox (base image) and once as SandboxPython (python image)
 // The actual image is determined by the container binding in wrangler.jsonc
 export { Sandbox };
-export { Sandbox as SandboxBase };
+export { Sandbox as SandboxPython };
 
 interface Env {
   Sandbox: DurableObjectNamespace<Sandbox>;
-  SandboxBase: DurableObjectNamespace<Sandbox>;
+  SandboxPython: DurableObjectNamespace<Sandbox>;
   TEST_BUCKET: R2Bucket;
   // R2 credentials for bucket mounting tests
   CLOUDFLARE_ACCOUNT_ID?: string;
@@ -66,10 +66,10 @@ export default {
     const keepAliveHeader = request.headers.get('X-Sandbox-KeepAlive');
     const keepAlive = keepAliveHeader === 'true';
 
-    // Select sandbox type: 'base' uses SandboxBase (no Python), anything else uses Sandbox (with Python)
+    // Select sandbox type: 'python' uses SandboxPython (with Python), anything else uses Sandbox (base, no Python)
     const sandboxType = request.headers.get('X-Sandbox-Type');
     const sandboxNamespace =
-      sandboxType === 'base' ? env.SandboxBase : env.Sandbox;
+      sandboxType === 'python' ? env.SandboxPython : env.Sandbox;
     const sandbox = getSandbox(sandboxNamespace, sandboxId, {
       keepAlive
     });