Merge pull request #13263 from mburke5678/eliminate-critical-pod-annotation

Fluent Pod Pre-emption and Priority

Author: mburke5678

modules/efk-logging-deploy-playbook.adoc

@@ -36,9 +36,25 @@ from the registry which can take up to a few minutes.
 +
 ----
 $ oc get pods -w
+logging-curator-1541129400-l5h77           0/1       Running   0          11h  <1>
+logging-es-data-master-ecu30lr4-1-deploy   0/1       Running   0          11h  <2>
+logging-fluentd-2lgwn                      1/1       Running   0          11h  <3>
+logging-fluentd-lmvms                      1/1       Running   0          11h
+logging-fluentd-p9nd7                      1/1       Running   0          11h
+logging-kibana-1-zk94k                     2/2       Running   0          11h  <4>
 ----
+<1> The Curator pod. Only one pod is needed for Curator.
+<2> The Elasticsearch pod on this host. 
+<3> The Fliuentd pods. There is one pod for each node in the cluster.
+<4> The Kibana pods.
 +
-When ready, the pod enter the *Running* status. 
+When ready, the pods enter the *Running* status. 
++
+[NOTE]
+====
+The logging-fluentd* pods are assigned a default priority class, `cluster-logging`, with a high priority that should prevent the pods from being descheduled.
+If you upgraded {product-title}, Fluentd with the `scheduler.alpha.kubernetes.io/critical-pod` annotation are automatically converted  to the `cluster-logging` priority class.
+====
 
 .Procedure