Add healthcheck endpoint for django-q2 clusters and scheduled tasks

cmsmanage/urls.py

@@ -25,6 +25,8 @@ from rest_framework import routers
 from membershipworks.api import router as membershipworks_router
 from paperwork.api import router as paperwork_router
 
+from . import views
+
 router = routers.DefaultRouter()
 router.registry.extend(paperwork_router.registry)
 router.registry.extend(membershipworks_router.registry)
@@ -59,6 +61,7 @@ urlpatterns = [
         ),
     ),
     path("api-auth/", include("rest_framework.urls")),
+    path("healthcheck", views.healthcheck),
     # path("markdownx/", include("markdownx.urls")),
 ]
 

cmsmanage/views.py

@@ -0,0 +1,84 @@
+import dataclasses
+import datetime
+from collections.abc import Iterable
+
+from django.http import HttpRequest, HttpResponse
+from django.utils import timezone
+
+from django_q.conf import Conf
+from django_q.models import Schedule, Task
+from django_q.status import Stat
+
+
+@dataclasses.dataclass
+class CheckResult:
+    message: str
+    ok: bool
+
+
+@dataclasses.dataclass
+class CheckResultOk(CheckResult):
+    message: str
+    ok: bool = True
+
+
+@dataclasses.dataclass
+class CheckResultFailure(CheckResult):
+    message: str
+    ok: bool = False
+
+
+def _check_clusters() -> CheckResult:
+    clusters = Stat.get_all()
+    happy_clusters = [
+        cluster for cluster in clusters if cluster.status in [Conf.IDLE, Conf.WORKING]
+    ]
+
+    if not len(clusters):
+        return CheckResultFailure("No clusters running!")
+    elif len(happy_clusters) != len(clusters):
+        return CheckResultFailure(f"{len(happy_clusters)}/{len(clusters)} clusters up")
+    else:
+        return CheckResultOk(f"{len(happy_clusters)}/{len(clusters)} clusters up")
+
+
+def _check_tasks() -> Iterable[CheckResult]:
+    now = timezone.now()
+    for schedule in Schedule.objects.all():
+        if not schedule.task:
+            yield CheckResultFailure(f"Scheduled task {schedule} has never run!")
+
+        else:
+            try:
+                task = Task.objects.get(id=schedule.task)
+            except Task.DoesNotExist:
+                yield CheckResultFailure(
+                    f"Scheduled task {schedule}'s last task doesn't exist!"
+                )
+                continue
+
+            if not task.success:
+                yield CheckResultFailure(
+                    f"Scheduled task {schedule} failed at {task.started}"
+                )
+            elif now - schedule.next_run > datetime.timedelta(hours=2):
+                yield CheckResultFailure(
+                    f"Scheduled task {schedule} stale, last run at {task.started}"
+                )
+            else:
+                yield CheckResultOk(
+                    f"Scheduled task {schedule} ok, last run at {task.started}"
+                )
+
+
+def healthcheck(request: HttpRequest):
+    checks: list[CheckResult] = [_check_clusters(), *_check_tasks()]
+
+    all_ok = all(check.ok for check in checks)
+    messages = (check.message for check in sorted(checks, key=lambda c: c.ok))
+
+    return HttpResponse(
+        ("OK: " if all_ok else "CRITICAL: ") + "\n".join(messages),
+        content_type="text/plain",
+        status=200 if all_ok else 500,
+    )