Add documentation for TaintNodesByCondition

docs/concepts/architecture/nodes.md

@@ -65,7 +65,27 @@ The node condition is represented as a JSON object. For example, the following r
 
 If the Status of the Ready condition is "Unknown" or "False" for longer than the `pod-eviction-timeout`, an argument is passed to the [kube-controller-manager](/docs/admin/kube-controller-manager) and all of the Pods on the node are scheduled for deletion by the Node Controller. The default eviction timeout duration is **five minutes**. In some cases when the node is unreachable, the apiserver is unable to communicate with the kubelet on it. The decision to delete the pods cannot be communicated to the kubelet until it re-establishes communication with the apiserver. In the meantime, the pods which are scheduled for deletion may continue to run on the partitioned node.
 
-In versions of Kubernetes prior to 1.5, the node controller would [force delete](/docs/concepts/workloads/pods/pod/#force-deletion-of-pods) these unreachable pods from the apiserver. However, in 1.5 and higher, the node controller does not force delete pods until it is confirmed that they have stopped running in the cluster. One can see these pods which may be running on an unreachable node as being in the "Terminating" or "Unknown" states. In cases where Kubernetes cannot deduce from the underlying infrastructure if a node has permanently left a cluster, the cluster administrator may need to delete the node object by hand.  Deleting the node object from Kubernetes causes all the Pod objects running on it to be deleted from the apiserver, freeing up their names.
+In versions of Kubernetes prior to 1.5, the node controller would [force delete](/docs/concepts/workloads/pods/pod/#force-deletion-of-pods)
+these unreachable pods from the apiserver. However, in 1.5 and higher, the node controller does not force delete pods until it is
+confirmed that they have stopped running in the cluster. One can see these pods which may be running on an unreachable node as being in
+the "Terminating" or "Unknown" states. In cases where Kubernetes cannot deduce from the underlying infrastructure if a node has
+permanently left a cluster, the cluster administrator may need to delete the node object by hand.  Deleting the node object from
+Kubernetes causes all the Pod objects running on it to be deleted from the apiserver, freeing up their names.
+
+Version 1.8 introduces an alpha feature that automatically creates
+[taints](/docs/concepts/configuration/taint-and-toleration) that represent conditions.
+To enable this behavior, pass an additional feature gate flag `--feature-gates=...,TaintNodesByCondition=true`
+to the API server, controller manager, and scheduler.
+When `TaintNodesByCondition` is enabled, the scheduler ignores conditions when considering a Node; instead
+it looks at the Node's taints and a Pod's tolerations.
+
+Now users can choose between the old scheduling model and a new, more flexible scheduling model.
+A Pod that does not have any tolerations gets scheduled according to the old model. But a Pod that 
+tolerates the taints of a particular Node can be scheduled on that Node.
+
+Note that because of small delay, usually less than one second, between time when condition is observed and a taint
+is created, it's possible that enabling this feature will slightly increase number of Pods that are successfully
+scheduled but rejected by the kubelet.
 
 ### Capacity
 
@@ -174,6 +194,9 @@ NodeController is responsible for adding taints corresponding to node problems l
 node unreachable or not ready. See [this documentation](/docs/concepts/configuration/taint-and-toleration)
 for details about `NoExecute` taints and the alpha feature.
 
+Starting in version 1.8, the node controller can be made responsible for creating taints that represent
+Node conditions. This is an alpha feature of version 1.8.
+
 ### Self-Registration of Nodes
 
 When the kubelet flag `--register-node` is true (the default), the kubelet will attempt to

docs/concepts/configuration/taint-and-toleration.md

@@ -188,7 +188,7 @@ running on the node as follows
 
 The above behavior is a beta feature. In addition, Kubernetes 1.6 has alpha
 support for representing node problems. In other words, the node controller
-automatically taints a node when certain condition is true. The builtin taints
+automatically taints a node when certain condition is true. The built-in taints
 currently include:
 
  * `node.alpha.kubernetes.io/notReady`: Node is not ready. This corresponds to
@@ -249,9 +249,20 @@ admission controller](https://git.k8s.io/kubernetes/plugin/pkg/admission/default
 
   * `node.alpha.kubernetes.io/unreachable`
   * `node.alpha.kubernetes.io/notReady`
-  * `node.kubernetes.io/memoryPressure`
-  * `node.kubernetes.io/diskPressure`
-  * `node.kubernetes.io/outOfDisk` (*only for critical pods*)
 
 This ensures that DaemonSet pods are never evicted due to these problems,
 which matches the behavior when this feature is disabled.
+
+## Taint Nodes by Condition
+
+Version 1.8 introduces an alpha feature that causes the node controller to create taints corresponding to
+Node conditions. When this feature is enabled, the scheduler does not check conditions; instead the scheduler checks taints. This assures that conditions don't affect what's scheduled onto the Node. The user can choose to ignore some of the Node's problems (represented as conditions) by adding appropriate Pod tolerations.
+
+To make sure that turning on this feature doesn't break DaemonSets, starting in version 1.8, the  DaemonSet controller automatically adds the following `NoSchedule` tolerations to all daemons:
+
+  * `node.kubernetes.io/memory-pressure`
+  * `node.kubernetes.io/disk-pressure`
+  * `node.kubernetes.io/out-of-disk` (*only for critical pods*)
+
+The above settings ensure backward compatibility, but we understand they may not fit all user's needs, which is why
+cluster admin may choose to add arbitrary tolerations to DaemonSets.

docs/concepts/workloads/controllers/daemonset.md

@@ -103,19 +103,22 @@ but they are created with `NoExecute` tolerations for the following taints with
 
  - `node.alpha.kubernetes.io/notReady`
  - `node.alpha.kubernetes.io/unreachable`
- - `node.alpha.kubernetes.io/memoryPressure`
- - `node.alpha.kubernetes.io/diskPressure`
-
-When the support to critical pods is enabled and the pods in a DaemonSet are
-labelled as critical, the Daemon pods are created with an additional
-`NoExecute` toleration for the `node.alpha.kubernetes.io/outOfDisk` taint with
-no `tolerationSeconds`.
 
 This ensures that when the `TaintBasedEvictions` alpha feature is enabled,
 they will not be evicted when there are node problems such as a network partition. (When the
 `TaintBasedEvictions` feature is not enabled, they are also not evicted in these scenarios, but
 due to hard-coded behavior of the NodeController rather than due to tolerations).
 
+ They also tolerate following `NoSchedule` taints:
+
+ - `node.kubernetes.io/memory-pressure`
+ - `node.kubernetes.io/disk-pressure`
+
+When the support to critical pods is enabled and the pods in a DaemonSet are
+labelled as critical, the Daemon pods are created with an additional
+`NoSchedule` toleration for the `node.kubernetes.io/out-of-disk` taint.
+
+Note that all above `NoSchedule` taints above are created only in version 1.8 or later if the alpha feature `TaintNodesByCondition` is enabled.
 
 ## Communicating with Daemon Pods