Add RemovePodsViolatingTopologySpreadConstraint strategy

This adds a strategy to balance pod topology domains based on the scheduler's
PodTopologySpread constraints. It attempts to find the minimum number of pods
that should be sent for eviction by comparing the largest domains in a topology
with the smallest domains in that topology.

README.md

@@ -37,6 +37,7 @@ Table of Contents
      * [RemovePodsViolatingInterPodAntiAffinity](#removepodsviolatinginterpodantiaffinity)
      * [RemovePodsViolatingNodeAffinity](#removepodsviolatingnodeaffinity)
      * [RemovePodsViolatingNodeTaints](#removepodsviolatingnodetaints)
+     * [RemovePodsViolatingTopologySpreadConstraint](#removepodsviolatingtopologyspreadconstraint)
      * [RemovePodsHavingTooManyRestarts](#removepodshavingtoomanyrestarts)
      * [PodLifeTime](#podlifetime)
   * [Filter Pods](#filter-pods)
@@ -102,17 +103,17 @@ See the [user guide](docs/user-guide.md) in the `/docs` directory.
 ## Policy and Strategies
 
 Descheduler's policy is configurable and includes strategies that can be enabled or disabled.
-Seven strategies `RemoveDuplicates`, `LowNodeUtilization`, `RemovePodsViolatingInterPodAntiAffinity`,
-`RemovePodsViolatingNodeAffinity`, `RemovePodsViolatingNodeTaints`, `RemovePodsHavingTooManyRestarts`, and `PodLifeTime`
-are currently implemented. As part of the policy, the parameters associated with the strategies can be configured too.
-By default, all strategies are enabled.
+Eight strategies `RemoveDuplicates`, `LowNodeUtilization`, `RemovePodsViolatingInterPodAntiAffinity`,
+`RemovePodsViolatingNodeAffinity`, `RemovePodsViolatingNodeTaints`, `RemovePodsViolatingTopologySpreadConstraint`,
+`RemovePodsHavingTooManyRestarts`, and `PodLifeTime` are currently implemented. As part of the policy, the
+parameters associated with the strategies can be configured too. By default, all strategies are enabled.
 
 The policy also includes common configuration for all the strategies:
 - `nodeSelector` - limiting the nodes which are processed
 - `evictLocalStoragePods` - allowing to evict pods with local storage
 - `maxNoOfPodsToEvictPerNode` - maximum number of pods evicted from each node (summed through all strategies)
 
-```
+```yaml
 apiVersion: "descheduler/v1alpha1"
 kind: "DeschedulerPolicy"
 nodeSelector: prod=dev
@@ -144,7 +145,7 @@ has any of these `Kind`s listed as an `OwnerRef`, that pod will not be considere
 |`thresholdPriorityClassName`|string (see [priority filtering](#priority-filtering))|
 
 **Example:**
-```
+```yaml
 apiVersion: "descheduler/v1alpha1"
 kind: "DeschedulerPolicy"
 strategies:
@@ -186,7 +187,7 @@ These thresholds, `thresholds` and `targetThresholds`, could be tuned as per you
 
 **Example:**
 
-```
+```yaml
 apiVersion: "descheduler/v1alpha1"
 kind: "DeschedulerPolicy"
 strategies:
@@ -236,7 +237,7 @@ node.
 
 **Example:**
 
-```
+```yaml
 apiVersion: "descheduler/v1alpha1"
 kind: "DeschedulerPolicy"
 strategies:
@@ -273,7 +274,7 @@ podA gets evicted from nodeA.
 
 **Example:**
 
-```
+```yaml
 apiVersion: "descheduler/v1alpha1"
 kind: "DeschedulerPolicy"
 strategies:
@@ -301,7 +302,7 @@ and will be evicted.
 
 **Example:**
 
-````
+````yaml
 apiVersion: "descheduler/v1alpha1"
 kind: "DeschedulerPolicy"
 strategies:
@@ -309,6 +310,31 @@ strategies:
     enabled: true
 ````
 
+### RemovePodsViolatingTopologySpreadConstraint
+
+This strategy makes sure that pods violating [topology spread constraints](https://kubernetes.io/docs/concepts/workloads/pods/pod-topology-spread-constraints/)
+are evicted from nodes. Specifically, it tries to evict the minimum number of pods required to balance topology domains to within each constraint's `maxSkew`.
+This strategy requires k8s version 1.18 at a minimum.
+
+**Parameters:**
+
+|Name|Type|
+|---|---|
+|`thresholdPriority`|int (see [priority filtering](#priority-filtering))|
+|`thresholdPriorityClassName`|string (see [priority filtering](#priority-filtering))|
+|`namespaces`|(see [namespace filtering](#namespace-filtering))|
+
+**Example:**
+
+```yaml
+apiVersion: "descheduler/v1alpha1"
+kind: "DeschedulerPolicy"
+strategies:
+  "RemovePodsViolatingTopologySpreadConstraint":
+     enabled: true
+```
+
+
 ### RemovePodsHavingTooManyRestarts
 
 This strategy makes sure that pods having too many restarts are removed from nodes. For example a pod with EBS/PD that 
@@ -328,7 +354,7 @@ which determines whether init container restarts should be factored into that ca
 
 **Example:**
 
-```
+```yaml
 apiVersion: "descheduler/v1alpha1"
 kind: "DeschedulerPolicy"
 strategies:
@@ -359,7 +385,7 @@ to `Running` and `Pending`.
 
 **Example:**
 
-```
+```yaml
 apiVersion: "descheduler/v1alpha1"
 kind: "DeschedulerPolicy"
 strategies:
@@ -383,10 +409,11 @@ The following strategies accept a `namespaces` parameter which allows to specify
 * `RemovePodsViolatingNodeAffinity`
 * `RemovePodsViolatingInterPodAntiAffinity`
 * `RemoveDuplicates`
+* `RemovePodsViolatingTopologySpreadConstraint`
 
 For example:
 
-```
+```yaml
 apiVersion: "descheduler/v1alpha1"
 kind: "DeschedulerPolicy"
 strategies:
@@ -404,7 +431,7 @@ strategies:
 In the examples `PodLifeTime` gets executed only over `namespace1` and `namespace2`.
 The similar holds for `exclude` field:
 
-```
+```yaml
 apiVersion: "descheduler/v1alpha1"
 kind: "DeschedulerPolicy"
 strategies:
@@ -432,7 +459,7 @@ is set to the value of `system-cluster-critical` priority class.
 E.g.
 
 Setting `thresholdPriority`
-```
+```yaml
 apiVersion: "descheduler/v1alpha1"
 kind: "DeschedulerPolicy"
 strategies:
@@ -445,7 +472,7 @@ strategies:
 ```
 
 Setting `thresholdPriorityClassName`
-```
+```yaml
 apiVersion: "descheduler/v1alpha1"
 kind: "DeschedulerPolicy"
 strategies: