Changes downloader alert to fire after 72h, not 21h (#1035)

The downloader used to download databases about every 8h. We have reconfigured
th downloader to only download about every 24h. Additionally, there is no need
for this alert to be super sensitive. It is okay if annotations happen with
databases that are only a couple days old. However, if downloader can't
download all the files for more than 3d, then fire an alert.

Author: nkinkade

config/federation/prometheus/alerts.yml

@@ -74,21 +74,18 @@ groups:
         the switch itself, or with the transit provider.
       dashboard: 'https://grafana.mlab-oti.measurementlab.net/d/SuqnZ6Hiz/?orgId=1&var-site_name={{$labels.site}}'
 
-# DownloaderIsFailingToUpdate: The downloader hasn't successfully retrieved the files in
-# at least 21 hours, meaning that at least the last two download attempts have failed.
+# DownloaderIsFailingToUpdate: The downloader hasn't successfully retrieved
+# all databases in more than 72 hours.
   - alert: DownloaderIsFailingToUpdate
-    expr: time() - downloader_last_success_time_seconds > (21 * 60 * 60)
+    expr: time() - downloader_last_success_time_seconds > (72 * 60 * 60)
     for: 1h
     labels:
       repo: dev-tracker
       severity: ticket
       cluster: prometheus-federation
     annotations:
-      summary: Neither of the last two attempts to download the maxmind or
-        routeviews feeds were successful.
-      description: Check for errors with the downloader service on grafana with
-        the downloader_error_count metric, or check the stackdriver logs for
-        the downloader cluster.
+      summary: The downloader hasn't successfully retrieved all databases in more than 3d
+      description: https://github.com/m-lab/ops-tracker/wiki/Alerts-&-Troubleshooting#downloaderisfailingtoupdate
       dashboard: https://grafana.mlab-oti.measurementlab.net/d/ZGuYht1mk/
 
 # DownloaderNotRunning: The downloader cluster crashed and not running at all.
@@ -1365,4 +1362,4 @@ groups:
     annotations:
       summary: Daily storage costs are 50% over the average storage costs for the month.
       description: https://github.com/m-lab/ops-tracker/wiki/Alerts-&-Troubleshooting#billing_dailystorageincrease
-      dashboard: https://grafana.mlab-oti.measurementlab.net/d/d8145875-e912-484e-b8f2-b77f63bd28a3/cloud-storage-usage?orgId=1
+      dashboard: https://grafana.mlab-oti.measurementlab.net/d/d8145875-e912-484e-b8f2-b77f63bd28a3/cloud-storage-usage?orgId=1