Merge pull request #101 from Novartis/delay_itemsource_init

Delay itemsource init

.github/workflows/pr-checks.yaml

@@ -52,17 +52,39 @@ jobs:
           eval "$(conda shell.bash hook)"
           conda activate cellxgene-gateway
           coverage report --fail-under 41
+          coverage report > coverage.txt
+          coverage html -i
           coverage xml -i
 
-      - name: "Upload coverage to Codecov"
-        if: ${{ github.event_name == 'push' || (github.event_name == 'pull_request' && github.event.pull_request.head.repo.full_name == github.repository) }}
-        uses: codecov/codecov-action@v1
+      - name: Upload coverage HTML report
+        uses: actions/upload-artifact@v4
         with:
-          token: ${{ secrets.CODECOV_TOKEN }}
-          files: ./coverage.xml
-          flags: unittests
-          env_vars: OS,PYTHON
-          name: codecov-umbrella
-          fail_ci_if_error: true
-          path_to_write_report: ./codecov_report.txt
-          verbose: true
+          name: coverage-html
+          path: htmlcov/
+          retention-days: 30
+
+      - name: Upload coverage xml
+        uses: actions/upload-artifact@v4
+        with:
+          name: coverage-xml
+          path: coverage.xml
+          retention-days: 30
+
+      - name: Upload coverage summary
+        uses: actions/upload-artifact@v4
+        with:
+          name: coverage-summary
+          path: coverage.txt
+          retention-days: 30
+      # - name: "Upload coverage to Codecov"
+      #   if: ${{ github.event_name == 'push' || (github.event_name == 'pull_request' && github.event.pull_request.head.repo.full_name == github.repository) }}
+      #   uses: codecov/codecov-action@v1
+      #   with:
+      #     token: ${{ secrets.CODECOV_TOKEN }}
+      #     files: ./coverage.xml
+      #     flags: unittests
+      #     env_vars: OS,PYTHON
+      #     name: codecov-umbrella
+      #     fail_ci_if_error: true
+      #     path_to_write_report: ./codecov_report.txt
+      #     verbose: true

README.md

@@ -115,6 +115,21 @@ docker run -it --rm \
 -p 8080:8080 \
 cellxgene-gateway
 ```
+## Running cellxgene gateway with start scripts
+
+For your convenience, we provide start scripts for flask, gunicorn and uwsgi.
+
+First, set up a .env
+```bash
+cp env_example .env
+# edit .env
+open .env
+```
+
+Then run the scripts in a subshell
+```bash
+( ./start_flask.sh )
+```
 
 # Customization
 

cellxgene_gateway/gateway.py

@@ -40,6 +40,12 @@ app = Flask(__name__)
 item_sources = []
 default_item_source = None
 
+# Guard for lazy initialization so tests can import this module without
+# triggering environment-dependent side effects. initialize_data_sources()
+# will set this to True when it has run.
+data_sources_initialized = False
+data_sources_init_lock = Lock()
+
 
 def _force_https(app):
     def wrapper(environ, start_response):
@@ -75,6 +81,46 @@ if (
         x_prefix=env.proxy_fix_prefix,
     )
 
+
+# WSGI middleware to ensure data sources are initialized before the first
+# WSGI request is handled. This guarantees initialization works under
+# Gunicorn/uWSGI (which import the module but don't call main()). The
+# initialize_data_sources() function is idempotent-protected by
+# data_sources_initialized and data_sources_init_lock.
+def _init_on_first_wsgi_request(wsgi_app):
+    def middleware(environ, start_response):
+        global data_sources_initialized
+        if not data_sources_initialized:
+            with data_sources_init_lock:
+                if not app.extensions.get("cellxgene_gateway", {}).get("launchtime"):
+                    app.extensions.setdefault("cellxgene_gateway", {})[
+                        "launchtime"
+                    ] = current_time_stamp()
+
+                if not data_sources_initialized:
+                    initialize_data_sources()
+
+                    env.validate()
+                    if not item_sources or not len(item_sources):
+                        raise Exception(
+                            "No data sources specified for Cellxgene Gateway"
+                        )
+
+                    global default_item_source
+                    if default_item_source is None:
+                        default_item_source = item_sources[0]
+
+                    data_sources_initialized = True
+        return wsgi_app(environ, start_response)
+
+    return middleware
+
+
+# Wrap the WSGI app so Gunicorn/uWSGI will trigger initialization when the
+# first request comes in. Tests that need initialization can call
+# initialize_data_sources() directly.
+app.wsgi_app = _init_on_first_wsgi_request(app.wsgi_app)
+
 cache = BackendCache()
 
 
@@ -207,7 +253,7 @@ entry_lock = Lock()
 
 
 def matching_source(source_name):
-    if source_name is None:
+    if source_name is None and default_item_source is not None:
         source_name = default_item_source.name
     matching = [i for i in item_sources if i.name == source_name]
     if len(matching) != 1:
@@ -254,6 +300,11 @@ def do_view(path, source_name=None):
             raise CellxgeneException("User not authorized to access this data", 403)
     elif match.status == CacheEntryStatus.error:
         raise ProcessException.from_cache_entry(match)
+    else:
+        raise CellxgeneException(
+            f"Unexpected cache entry status {match.status} for key {match.key.descriptor}",
+            500,
+        )
 
 
 @app.route("/cache_status", methods=["GET"])
@@ -267,28 +318,40 @@ def do_GET_status():
 
 @app.route("/cache_status.json", methods=["GET"])
 def do_GET_status_json():
+    def map_entry(entry):
+        dataset = entry.key.h5ad_item.descriptor
+        annotation_file = entry.key.annotation_descriptor
+        return {
+            "dataset": dataset,
+            "annotation_file": annotation_file,
+            "launchtime": entry.launchtime,
+            "last_access": entry.timestamp,
+            "status": entry.status.name,
+        }
+
     return json.dumps(
         {
-            "launchtime": app.launchtime,
-            "entry_list": [
-                {
-                    "dataset": entry.key.dataset,
-                    "annotation_file": entry.key.annotation_file,
-                    "launchtime": entry.launchtime,
-                    "last_access": entry.timestamp,
-                    "status": entry.status,
-                }
-                for entry in cache.entry_list
-            ],
+            "launchtime": app.extensions.get("cellxgene_gateway", {}).get("launchtime"),
+            "entry_list": [map_entry(entry) for entry in cache.entry_list],
         }
     )
 
 
-@app.route("/relaunch/<path:path>", methods=["GET"])
-def do_relaunch(path):
-    source_name = request.args.get("source_name") or default_item_source.name
+def get_cache_key(path):
+    if request.args.get("source_name"):
+        source_name = request.args.get("source_name")
+    elif default_item_source:
+        source_name = default_item_source.name
+    else:
+        source_name = None
     source = matching_source(source_name)
     key = CacheKey.for_lookup(source, source.lookup(path))
+    return key
+
+
+@app.route("/relaunch/<path:path>", methods=["GET"])
+def do_relaunch(path):
+    key = get_cache_key(path)
     match = cache.check_entry(key)
     if not match is None:
         match.terminate()
@@ -300,9 +363,7 @@ def do_relaunch(path):
 
 @app.route("/terminate/<path:path>", methods=["GET"])
 def do_terminate(path):
-    source_name = request.args.get("source_name") or default_item_source.name
-    source = matching_source(source_name)
-    key = CacheKey.for_lookup(source, source.lookup(path))
+    key = get_cache_key(path)
     match = cache.check_entry(key)
     if not match is None:
         match.terminate()
@@ -315,28 +376,25 @@ def ip_address():
     return set_no_cache(resp)
 
 
-def launch():
-    env.validate()
-    if not item_sources or not len(item_sources):
-        raise Exception("No data sources specified for Cellxgene Gateway")
-
-    global default_item_source
-    if default_item_source is None:
-        default_item_source = item_sources[0]
-
+def start_pruner_thread():
     pruner = PruneProcessCache(cache)
-
-    background_thread = Thread(target=pruner)
+    # Run the pruner as a daemon thread so it won't block interpreter
+    # shutdown (for example when Ctrl-C is used in the main thread).
+    # This avoids "Exception ignored in: <module 'threading'...>" at exit.
+    background_thread = Thread(target=pruner, daemon=True)
     background_thread.start()
 
-    app.launchtime = current_time_stamp()
+
+def launch():
+    start_pruner_thread()
+
+    app.extensions.setdefault("cellxgene_gateway", {})[
+        "launchtime"
+    ] = current_time_stamp()
     app.run(host="0.0.0.0", port=env.gateway_port, debug=False)
 
 
-# When using servers like Gunicorn or uWSGI, this file is imported rather than run directly.
-# As a result, the main() function is never called automatically.
-# Therefore, we must initialize the data sources at import time to ensure they are available.
-initialize_data_sources()
+app.extensions.setdefault("cellxgene_gateway", {})["launchtime"] = None
 
 
 def main():

env_example

@@ -0,0 +1,3 @@
+export CELLXGENE_LOCATION=$(pwd)/.venv/bin/cellxgene
+export CELLXGENE_DATA=../cellxgene_data
+export GATEWAY_IP=127.0.0.1

run.sh.example

@@ -1,6 +0,0 @@
-export CELLXGENE_LOCATION=$(pwd)/.cellxgene-gateway/bin/cellxgene
-export CELLXGENE_DATA=../cellxgene_data
-export GATEWAY_IP=127.0.0.1
-
-#Once these are set, you run like a normal Flask app
-cellxgene-gateway

start_flask.sh

@@ -0,0 +1,41 @@
+#!/bin/bash
+
+# start_gunicorn.sh - Start Cellxgene Gateway with Gunicorn
+#
+# PREREQUISITES:
+# - Gunicorn installed (included with cellxgene 1.3.0, or: pip install gunicorn)
+# - Virtual environment activated
+# - .env file with CELLXGENE_LOCATION and CELLXGENE_DATA (or CELLXGENE_BUCKET)
+#
+# USAGE:
+# ./start_gunicorn.sh
+
+# Exit on error
+set -e
+
+# Get the directory where this script is located
+SCRIPT_DIR="$( cd "$( dirname "${BASH_SOURCE[0]}" )" && pwd )"
+
+
+# Source environment variables
+echo "Loading environment variables..."
+if [ -f "$SCRIPT_DIR/.env" ]; then
+    source "$SCRIPT_DIR/.env"
+else
+    echo "Error: .env file not found at $SCRIPT_DIR/.env"
+    echo "Please create it with required environment variables"
+    exit 1
+fi
+
+# Verify required environment variables
+if [ -z "$CELLXGENE_LOCATION" ]; then
+    echo "Error: CELLXGENE_LOCATION not set"
+    exit 1
+fi
+
+if [ -z "$CELLXGENE_DATA" ] && [ -z "$CELLXGENE_BUCKET" ]; then
+    echo "Error: Either CELLXGENE_DATA or CELLXGENE_BUCKET must be set"
+    exit 1
+fi
+
+exec cellxgene-gateway

start_gunicorn.sh

@@ -0,0 +1,91 @@
+#!/bin/bash
+
+# start_gunicorn.sh - Start Cellxgene Gateway with Gunicorn
+#
+# PREREQUISITES:
+# - Gunicorn installed (included with cellxgene 1.3.0, or: pip install gunicorn)
+# - Virtual environment activated
+# - .env file with CELLXGENE_LOCATION and CELLXGENE_DATA (or CELLXGENE_BUCKET)
+#
+# USAGE:
+# ./start_gunicorn.sh
+
+# Exit on error
+set -e
+
+# Get the directory where this script is located
+SCRIPT_DIR="$( cd "$( dirname "${BASH_SOURCE[0]}" )" && pwd )"
+
+# Source environment variables
+echo "Loading environment variables..."
+if [ -f "$SCRIPT_DIR/.env" ]; then
+    source "$SCRIPT_DIR/.env"
+else
+    echo "Error: .env file not found at $SCRIPT_DIR/.env"
+    echo "Please create it with required environment variables"
+    exit 1
+fi
+
+# Verify required environment variables
+if [ -z "$CELLXGENE_LOCATION" ]; then
+    echo "Error: CELLXGENE_LOCATION not set"
+    exit 1
+fi
+
+if [ -z "$CELLXGENE_DATA" ] && [ -z "$CELLXGENE_BUCKET" ]; then
+    echo "Error: Either CELLXGENE_DATA or CELLXGENE_BUCKET must be set"
+    exit 1
+fi
+
+# Gunicorn configuration
+# WARNING: Multi-worker mode has cache synchronization issues (see plans/002-shared-cache-implementation.md)
+# Each worker maintains its own in-memory cache, causing 404s for static assets when different
+# workers handle requests for the same dataset. Use GUNICORN_WORKERS=1 until shared cache is implemented.
+WORKERS=${GUNICORN_WORKERS:-1}
+BIND=${GATEWAY_IP:-0.0.0.0}:${GATEWAY_PORT:-5005}
+TIMEOUT=${GUNICORN_TIMEOUT:-120}
+WORKER_CLASS=${GUNICORN_WORKER_CLASS:-sync}
+KEEPALIVE=${GUNICORN_KEEPALIVE:-5}
+LOG_LEVEL=${GUNICORN_LOG_LEVEL:-info}
+
+# Production optimization: enable backed mode to reduce memory usage
+export GATEWAY_ENABLE_BACKED_MODE=${GATEWAY_ENABLE_BACKED_MODE:-true}
+
+# Check if gunicorn is installed
+if ! command -v gunicorn &> /dev/null; then
+    echo "Error: gunicorn not found. Install with: pip install gunicorn"
+    exit 1
+fi
+
+# Display configuration
+echo "Starting Cellxgene Gateway with Gunicorn..."
+echo "Configuration:"
+echo "  Data source: ${CELLXGENE_DATA:-$CELLXGENE_BUCKET}"
+echo "  Binding to: $BIND"
+echo "  Workers: $WORKERS"
+echo "  Worker class: $WORKER_CLASS"
+echo "  Timeout: ${TIMEOUT}s"
+echo "  Keepalive: ${KEEPALIVE}s"
+echo "  Log level: $LOG_LEVEL"
+echo "  Backed mode: ${GATEWAY_ENABLE_BACKED_MODE}"
+echo ""
+
+cd "$SCRIPT_DIR"
+
+# Start Gunicorn with optimized settings
+# Additional options you can add via environment variables:
+# - GUNICORN_MAX_REQUESTS: Restart worker after N requests (prevents memory leaks)
+# - GUNICORN_MAX_REQUESTS_JITTER: Add randomness to max-requests
+exec gunicorn cellxgene_gateway.gateway:app \
+    --workers "$WORKERS" \
+    --worker-class "$WORKER_CLASS" \
+    --bind "$BIND" \
+    --timeout "$TIMEOUT" \
+    --keep-alive "$KEEPALIVE" \
+    --access-logfile - \
+    --error-logfile - \
+    --log-level "$LOG_LEVEL" \
+    --preload \
+    ${GUNICORN_MAX_REQUESTS:+--max-requests "$GUNICORN_MAX_REQUESTS"} \
+    ${GUNICORN_MAX_REQUESTS_JITTER:+--max-requests-jitter "$GUNICORN_MAX_REQUESTS_JITTER"} \
+    "$@"

start_uwsgi.sh

@@ -0,0 +1,89 @@
+#!/bin/bash
+
+# start_uwsgi.sh - Start Cellxgene Gateway with uWSGI
+#
+# PREREQUISITES:
+# - uWSGI installed (pip install uwsgi)
+# - Virtual environment activated
+# - .env file with CELLXGENE_LOCATION and CELLXGENE_DATA (or CELLXGENE_BUCKET)
+#
+# USAGE:
+# ./start_uwsgi.sh
+
+# Exit on error
+set -e
+
+# Get the directory where this script is located
+SCRIPT_DIR="$( cd "$( dirname "${BASH_SOURCE[0]}" )" && pwd )"
+
+# Source environment variables
+echo "Loading environment variables..."
+if [ -f "$SCRIPT_DIR/.env" ]; then
+    source "$SCRIPT_DIR/.env"
+else
+    echo "Error: .env file not found at $SCRIPT_DIR/.env"
+    echo "Please create it with required environment variables"
+    exit 1
+fi
+
+# Verify required environment variables
+if [ -z "$CELLXGENE_LOCATION" ]; then
+    echo "Error: CELLXGENE_LOCATION not set"
+    exit 1
+fi
+
+if [ -z "$CELLXGENE_DATA" ] && [ -z "$CELLXGENE_BUCKET" ]; then
+    echo "Error: Either CELLXGENE_DATA or CELLXGENE_BUCKET must be set"
+    exit 1
+fi
+
+# uWSGI configuration
+# WARNING: Multi-worker mode has cache synchronization issues (see plans/002-shared-cache-implementation.md)
+# Each worker maintains its own in-memory cache, causing 404s for static assets when different
+# workers handle requests for the same dataset. Use UWSGI_WORKERS=1 until shared cache is implemented.
+WORKERS=${UWSGI_WORKERS:-1}
+HOST=${GATEWAY_IP:-0.0.0.0}
+PORT=${GATEWAY_PORT:-5005}
+TIMEOUT=${UWSGI_TIMEOUT:-120}
+THREADS=${UWSGI_THREADS:-1}
+
+# Production optimization: enable backed mode to reduce memory usage
+export GATEWAY_ENABLE_BACKED_MODE=${GATEWAY_ENABLE_BACKED_MODE:-true}
+
+# Check if uwsgi is installed
+if ! command -v uwsgi &> /dev/null; then
+    echo "Error: uwsgi not found. Install with: pip install uwsgi"
+    exit 1
+else
+    # Display configuration
+    echo "Starting Cellxgene Gateway with uWSGI..."
+    echo "Configuration:"
+    echo "  Data source: ${CELLXGENE_DATA:-$CELLXGENE_BUCKET}"
+    echo "  Binding to: $HOST:$PORT"
+    echo "  Workers: $WORKERS"
+    echo "  Threads: $THREADS"
+    echo "  Timeout: ${TIMEOUT}s"
+    echo "  Backed mode: ${GATEWAY_ENABLE_BACKED_MODE}"
+    echo ""
+
+    cd "$SCRIPT_DIR"
+
+    # Start uWSGI with optimized settings
+    # Additional options you can add via environment variables:
+    # - UWSGI_MAX_REQUESTS: Restart worker after N requests (prevents memory leaks)
+    exec uwsgi \
+        --http "$HOST:$PORT" \
+        --module cellxgene_gateway.gateway:app \
+        --workers "$WORKERS" \
+        --threads "$THREADS" \
+        --harakiri "$TIMEOUT" \
+        --master \
+        --enable-threads \
+        --single-interpreter \
+        --need-app \
+        --die-on-term \
+        --log-x-forwarded-for \
+        ${UWSGI_MAX_REQUESTS:+--max-requests "$UWSGI_MAX_REQUESTS"} \
+        "$@"
+fi
+

tests/items/s3/test_s3item_source.py

@@ -8,24 +8,16 @@ from unittest.mock import MagicMock, Mock, patch
 from cellxgene_gateway.items.item import ItemType
 from cellxgene_gateway.items.s3.s3item import S3Item
 from cellxgene_gateway.items.s3.s3item_source import S3ItemSource
+from cellxgene_gateway.gateway import app
 
 
 class TestScanDirectory(unittest.TestCase):
     def setUp(self):
-        self._tmpdir = tempfile.mkdtemp()
-        self._cellxgene_data = os.environ.get("CELLXGENE_DATA", "")
-        os.environ["CELLXGENE_DATA"] = self._tmpdir
-
-        from cellxgene_gateway.gateway import app
 
         self.app = app
 
     def tearDown(self):
-        if self._cellxgene_data:
-            os.environ["CELLXGENE_DATA"] = self._cellxgene_data
-        else:
-            del os.environ["CELLXGENE_DATA"]
-        shutil.rmtree(self._tmpdir)
+        pass
 
     @patch("s3fs.S3FileSystem")
     def test_GIVEN_invalid_bucket_THEN_throws_error(self, s3func):

tests/test_cache_entry.py

@@ -10,6 +10,7 @@ from cellxgene_gateway.cache_key import CacheKey
 from cellxgene_gateway.items.item import ItemType
 from cellxgene_gateway.items.file.fileitem import FileItem
 from cellxgene_gateway.items.file.fileitem_source import FileItemSource
+from cellxgene_gateway.gateway import app
 
 key = CacheKey(
     FileItem("/czi/", name="pbmc3k.h5ad", type=ItemType.h5ad),
@@ -19,12 +20,6 @@ key = CacheKey(
 
 class TestRenderEntry(unittest.TestCase):
     def setUp(self):
-        self._tmpdir = tempfile.mkdtemp()
-        self._cellxgene_data = os.environ.get("CELLXGENE_DATA", "")
-        os.environ["CELLXGENE_DATA"] = self._tmpdir
-
-        from cellxgene_gateway.gateway import app
-
         self.app = app
         self.app_context = self.app.test_request_context()
         self.app_context.push()
@@ -32,11 +27,6 @@ class TestRenderEntry(unittest.TestCase):
 
     def tearDown(self):
         self.app_context.pop()
-        if self._cellxgene_data:
-            os.environ["CELLXGENE_DATA"] = self._cellxgene_data
-        else:
-            del os.environ["CELLXGENE_DATA"]
-        shutil.rmtree(self._tmpdir)
 
     def test_GIVEN_key_and_port_THEN_returns_loading_CacheEntry(self):
         entry = CacheEntry.for_key("some-key", 1)

tests/test_filecrawl.py

@@ -13,6 +13,7 @@ from cellxgene_gateway.filecrawl import (
 from cellxgene_gateway.items.file.fileitem import FileItem
 from cellxgene_gateway.items.file.fileitem_source import FileItemSource
 from cellxgene_gateway.items.item import ItemTree, ItemType
+from cellxgene_gateway.gateway import app
 
 source = FileItemSource("/tmp")
 
@@ -29,23 +30,12 @@ def make_entry(subpath="somepath", annotations=None):
 
 class TestRenderEntry(unittest.TestCase):
     def setUp(self):
-        self._tmpdir = tempfile.mkdtemp()
-        self._cellxgene_data = os.environ.get("CELLXGENE_DATA", "")
-        os.environ["CELLXGENE_DATA"] = self._tmpdir
-
-        from cellxgene_gateway.gateway import app
-
         self.app = app
         self.app_context = self.app.test_request_context()
         self.app_context.push()
 
     def tearDown(self):
         self.app_context.pop()
-        if self._cellxgene_data:
-            os.environ["CELLXGENE_DATA"] = self._cellxgene_data
-        else:
-            del os.environ["CELLXGENE_DATA"]
-        shutil.rmtree(self._tmpdir)
 
     def test_GIVEN_path_both_slash_THEN_view_has_single_slash(self):
         entry = make_entry(subpath="/somepath/")
@@ -71,23 +61,12 @@ class TestRenderEntry(unittest.TestCase):
 class TestRenderAnnotation(unittest.TestCase):
 
     def setUp(self):
-        self._tmpdir = tempfile.mkdtemp()
-        self._cellxgene_data = os.environ.get("CELLXGENE_DATA", "")
-        os.environ["CELLXGENE_DATA"] = self._tmpdir
-
-        from cellxgene_gateway.gateway import app
-
         self.app = app
         self.app_context = self.app.test_request_context()
         self.app_context.push()
 
     def tearDown(self):
         self.app_context.pop()
-        if self._cellxgene_data:
-            os.environ["CELLXGENE_DATA"] = self._cellxgene_data
-        else:
-            del os.environ["CELLXGENE_DATA"]
-        shutil.rmtree(self._tmpdir)
 
     @patch("cellxgene_gateway.filecrawl.enable_annotations", new=True)
     def test_GIVEN_no_annotation_THEN_new_alone(self):
@@ -145,23 +124,12 @@ class TestRenderItemSource(unittest.TestCase):
 
 class TestRenderItemTree(unittest.TestCase):
     def setUp(self):
-        self._tmpdir = tempfile.mkdtemp()
-        self._cellxgene_data = os.environ.get("CELLXGENE_DATA", "")
-        os.environ["CELLXGENE_DATA"] = self._tmpdir
-
-        from cellxgene_gateway.gateway import app
-
         self.app = app
         self.app_context = self.app.test_request_context()
         self.app_context.push()
 
     def tearDown(self):
         self.app_context.pop()
-        if self._cellxgene_data:
-            os.environ["CELLXGENE_DATA"] = self._cellxgene_data
-        else:
-            del os.environ["CELLXGENE_DATA"]
-        shutil.rmtree(self._tmpdir)
 
     @patch("cellxgene_gateway.items.file.fileitem_source.FileItemSource")
     def test_GIVEN_deep_nested_dirs_THEN_includes_dirs_in_output(self, item_source):

tests/test_gateway_status_json.py

@@ -0,0 +1,54 @@
+import json
+import unittest
+from types import SimpleNamespace
+
+from cellxgene_gateway.gateway import do_GET_status_json, app, cache
+from cellxgene_gateway.cache_entry import CacheEntry, CacheEntryStatus
+
+
+class TestGatewayStatusJson(unittest.TestCase):
+    def test_do_GET_status_json_returns_expected_structure(self):
+        # Create a minimal fake key with required attributes
+        h5ad_item = SimpleNamespace(descriptor="somedir/dataset.h5ad")
+        key = SimpleNamespace(
+            h5ad_item=h5ad_item,
+            annotation_descriptor="somedir/dataset_annotations/foo.csv",
+        )
+
+        # Create a CacheEntry with known launchtime/timestamp/status
+        entry = CacheEntry(
+            None,
+            key,
+            8000,
+            111,
+            222,
+            CacheEntryStatus.loaded,
+            None,
+            None,
+            None,
+            None,
+        )
+
+        # Install into the gateway cache and set app launchtime
+        cache.entry_list = [entry]
+        app.extensions.setdefault("cellxgene_gateway", {})["launchtime"] = "LAUNCH_TIME"
+
+        rv = do_GET_status_json()
+
+        data = json.loads(rv)
+        # top-level launchtime comes from app.extensions
+        self.assertEqual("LAUNCH_TIME", data["launchtime"])
+
+        self.assertIn("entry_list", data)
+        self.assertEqual(1, len(data["entry_list"]))
+
+        e = data["entry_list"][0]
+        self.assertEqual("somedir/dataset.h5ad", e["dataset"])
+        self.assertEqual("somedir/dataset_annotations/foo.csv", e["annotation_file"])
+        self.assertEqual("loaded", e["status"])
+        self.assertEqual(111, e["launchtime"])
+        self.assertEqual(222, e["last_access"])
+
+
+if __name__ == "__main__":
+    unittest.main()