Spread out backfill jobs to keep resource usage more stable

class/defaults.yml

@@ -13,9 +13,6 @@ parameters:
         repository: appuio/appuio-reporting
         tag: v0.2.1
 
-    schedules:
-      backfill: '15 * * * *'
-
     schedules_suspend:
       backfill: false
 

component/main.jsonnet

@@ -183,10 +183,10 @@ local backfillCJ = function(rule, product, index)
       },
       // Keeping infinite jobs is not possible. Keep at least one month worth of jobs.
       failedJobsHistoryLimit: 24 * 32,
+      schedule: '%s * * * *' % (index % 41+10) # spread out the jobs over the hour, from 10 to 50 minutes
     },
   };
 
-
 // Define outputs below
 {
   '00_namespace': kube.Namespace(params.namespace) {

docs/modules/ROOT/pages/references/parameters.adoc

@@ -11,36 +11,38 @@ default:: `${_instance}`
 The namespace in which to deploy this component.
 
 
-== `images`
+== `namespaceMetadata`
 
 [horizontal]
 type:: dictionary
-default:: https://github.com/appuio/component-appuio-reporting/blob/master/class/defaults.yml[See `class/defaults.yml`].
+default::
++
+[source,yaml]
+----
+namespaceMetadata:
+  labels:
+    openshift.io/cluster-monitoring: 'true'
+----
 
-Dictionary containing the container images used by this component.
+The namespace metadata to apply to the namespace created by this component.
 
 
-=== `images.reporting`
+== `images`
 
 [horizontal]
 type:: dictionary
+default:: https://github.com/appuio/component-appuio-reporting/blob/master/class/defaults.yml[See `class/defaults.yml`].
 
-The image used for managing data in the reporting database.
-
+Dictionary containing the container images used by this component.
 
 
-== `schedules`
+=== `images.reporting`
 
 [horizontal]
 type:: dictionary
-default::
-+
-[source,yaml]
-----
-backfill: '15 * * * *'
-----
 
-Dictionary managing the schedules of the periodic jobs (implemented as Kubernetes CronJobs).
+The image used for managing data in the reporting database.
+
 
 == `development_mode`
 
@@ -239,6 +241,7 @@ The jsonnet parameter takes precedence over the pattern parameter if both are sp
 
 Note that there is no jsonnet variant for the `query_pattern` parameter.
 
+CronJobs are spread across the hour to allow for a more even load on the cluster and Prometheus.
 
 
 == `network_policies.target_namespaces`