datavis: log detailed errors for outdated statistics (#85742)

combo/apps/dataviz/utils.py

@@ -1,18 +1,24 @@
+import datetime
 import json
+import logging
 
 from django.conf import settings
+from django.utils.timezone import now
 from requests.exceptions import RequestException
 
 from combo.utils import requests
 
 from .models import Statistic
 
+logger = logging.getLogger('combo.apps.dataviz')
+
 
 def update_available_statistics():
     if not settings.KNOWN_SERVICES:
         return
 
     results = []
+    temporary_unavailable_sites = []
     for provider in settings.STATISTICS_PROVIDERS:
         if isinstance(provider, dict):
             url = provider['url']
@@ -35,6 +41,7 @@ def update_available_statistics():
                 )
                 response.raise_for_status()
             except RequestException:
+                temporary_unavailable_sites.append((provider, site_key))
                 continue
 
             try:
@@ -93,4 +100,23 @@ def update_available_statistics():
         available_stats = available_stats.exclude(
             slug=stat.slug, site_slug=stat.site_slug, service_slug=stat.service_slug
         )
+
+    # set last_update on all seen statistics
+    Statistic.objects.exclude(pk__in=available_stats).update(last_update=now())
+
+    for service_slug, site_slug in temporary_unavailable_sites:
+        available_stats = available_stats.exclude(site_slug=site_slug, service_slug=service_slug)
     available_stats.update(available=False)
+
+    # log errors for outdated statistics
+    sites_with_outdated_statistics = set()
+    outdated_hours = 48
+    for available_stat in Statistic.objects.filter(available=True):
+        time_since_last_update = now() - available_stat.last_update
+        if time_since_last_update > datetime.timedelta(hours=outdated_hours):
+            sites_with_outdated_statistics.add(available_stat.site_title)
+
+    for title in sites_with_outdated_statistics:
+        logger.error(
+            f'statistics from "{title}" have not been available for more than %s hours.', outdated_hours
+        )

tests/test_dataviz.py

@@ -15,6 +15,7 @@ from httmock import HTTMock, remember_called, urlmatch, with_httmock
 from requests.exceptions import HTTPError
 
 from combo.apps.dataviz.models import ChartFiltersCell, ChartNgCell, Gauge, Statistic, UnsupportedDataSet
+from combo.apps.dataviz.utils import update_available_statistics
 from combo.data.models import Page, ValidityInfo
 from combo.utils.spooler import refresh_statistics_data
 
@@ -3451,3 +3452,40 @@ def test_chart_filters_cell_required_boolean(new_api_statistics, app, admin_user
     # different value, boolean filter is hidden
     resp = app.get('/')
     assert not 'filter-test' in resp.form.fields
+
+
+def test_update_available_statistics(app, settings, caplog, freezer):
+    caplog.set_level('WARNING')
+
+    settings.KNOWN_SERVICES = {
+        'authentic': {
+            'connection': {
+                'title': 'Connection',
+                'url': 'https://authentic.example.com',
+                'secret': 'combo',
+                'orig': 'combo',
+            }
+        }
+    }
+    settings.STATISTICS_PROVIDERS = ['authentic']
+
+    assert Statistic.objects.count() == 0
+
+    with HTTMock(new_api_mock):
+        update_available_statistics()
+    assert Statistic.objects.count() == len(STATISTICS_LIST['data'])
+    assert not caplog.messages
+
+    # move 47 hours in the future, still no error
+    freezer.move_to(datetime.timedelta(hours=47))
+    update_available_statistics()
+    assert caplog.messages
+    assert all(record.levelname == 'WARNING' for record in caplog.records)
+    caplog.clear()
+    caplog.set_level('ERROR')
+
+    # move 2 hours later, an error is logged
+    freezer.move_to(datetime.timedelta(hours=2))
+    update_available_statistics()
+    assert caplog.records
+    assert caplog.messages == ['statistics from "Connection" have not been available for more than 48 hours.']