Add healthcheck task

Author: peter-rauscher

.env.template

@@ -29,3 +29,10 @@ SOLESEARCH_STOCKX_CALLBACK_URL=
 SOLESEARCH_STOCKX_CALLBACK_STATE=
 SOLESEARCH_SESSION_SECRET=
 # ===================
+
+# === AWS ===
+AWS_HEALTHCHECK_URL=
+AWS_ACCESS_KEY_ID=
+AWS_SECRET_ACCESS_KEY=
+AWS_REGION=
+# ===============

poetry.lock

@@ -21,6 +21,8 @@ redis = "^5.2.0"
 logfire = { extras = ["fastapi", "celery", "sqlalchemy"], version = "^3.4.0" }
 beautifulsoup4 = "^4.13.3"
 debugpy = "^1.8.13"
+boto3 = "^1.37.13"
+requests-aws4auth = "^1.3.1"
 
 [tool.poetry.group.dev.dependencies]
 ruff = "^0.9.4"

pyproject.toml

@@ -20,6 +20,20 @@
 if not ENVIRONMENT:
     raise OSError("ENVIRONMENT environment variable not set.")
 
+AWS_HEALTHCHECK_URL = os.environ.get("AWS_HEALTHCHECK_URL")
+AWS_ACCESS_KEY_ID = os.environ.get("AWS_ACCESS_KEY_ID")
+AWS_SECRET_ACCESS_KEY = os.environ.get("AWS_SECRET_ACCESS_KEY")
+AWS_REGION = os.environ.get("AWS_REGION", "us-east-1")
+
+if not AWS_HEALTHCHECK_URL:
+    raise OSError("AWS_HEALTHCHECK_URL environment variable not set.")
+
+if not AWS_ACCESS_KEY_ID:
+    raise OSError("AWS_ACCESS_KEY_ID environment variable not set.")
+
+if not AWS_SECRET_ACCESS_KEY:
+    raise OSError("AWS_SECRET_ACCESS_KEY environment variable not set.")
+
 DATA_DIR = "/var/data/solesearch"
 HTML_DIR = os.path.join(DATA_DIR, "html")
 JSON_DIR = os.path.join(DATA_DIR, "json")

src/solesearch_api/config.py

@@ -28,4 +28,28 @@ def init_beat(*args, **kwargs):
     config_source=CeleryConfig,
 )
 
+app.conf.beat_schedule = {
+    "adidas-new-releases-every-4-hours": {
+        "task": "solesearch_api.tasks.scraping.retail.adidas.new_releases",
+        "schedule": 4 * 60 * 60,  # every 4 hours
+    },
+    "adidas-pdp-every-4-hours": {
+        "task": "solesearch_api.tasks.scraping.retail.adidas.pdp",
+        "schedule": 4 * 60 * 60,  # every 4 hours
+    },
+    "nike-new-releases-every-4-hours": {
+        "task": "solesearch_api.tasks.scraping.retail.nike.new_releases",
+        "schedule": 4 * 60 * 60,  # every 4 hours
+    },
+    "nike-in-stock-every-4-hours": {
+        "task": "solesearch_api.tasks.scraping.retail.nike.in_stock",
+        "schedule": 4 * 60 * 60,  # every 4 hours
+    },
+    "healthcheck-every-minute": {
+        "task": "solesearch_api.tasks.alerting.healthcheck",
+        "schedule": 60 * 10,  # every 10 minutes
+    },
+}
+
 import solesearch_api.tasks.scraping
+import solesearch_api.tasks.alerting.healthcheck

src/solesearch_api/tasks/__init__.py

@@ -0,0 +1,30 @@
+import logging
+import requests
+from requests_aws4auth import AWS4Auth
+
+from solesearch_api.config import (
+    AWS_ACCESS_KEY_ID,
+    AWS_HEALTHCHECK_URL,
+    AWS_SECRET_ACCESS_KEY,
+    AWS_REGION,
+)
+from solesearch_api.tasks import app
+
+logger = logging.getLogger(__name__)
+
+auth = AWS4Auth(
+    AWS_ACCESS_KEY_ID,
+    AWS_SECRET_ACCESS_KEY,
+    AWS_REGION,
+    "execute-api",
+)
+
+
+@app.task(name="solesearch_api.tasks.alerting.healthcheck")
+def healthcheck():
+    response = requests.post(
+        AWS_HEALTHCHECK_URL,
+        json={"ping": True},
+        auth=auth,
+    )
+    response.raise_for_status()