add remove too many restarts policy

.gitignore

@@ -0,0 +1,2 @@
+_output/
+.idea/

README.md

@@ -253,6 +253,23 @@ strategies:
       - "requiredDuringSchedulingIgnoredDuringExecution"
 ```
 
+
+### RemovePodsHavingTooManyRestarts
+
+This strategy makes sure that pods having too many restarts are removed from nodes. For example a pod with EBS/PD that can't get the volume/disk attached to the instance, then the pod should be re-scheduled to other nodes.
+
+```
+apiVersion: "descheduler/v1alpha1"
+kind: "DeschedulerPolicy"
+strategies:
+  "RemovePodsHavingTooManyRestarts":
+     enabled: true
+     params:
+       podsHavingTooManyRestarts:
+         podeRestartThresholds: 100
+         includingInitContainers: true
+```
+
 ## Pod Evictions
 
 When the descheduler decides to evict pods from a node, it employs following general mechanism:

examples/policy.yaml

@@ -17,3 +17,9 @@ strategies:
            "cpu" : 50
            "memory": 50
            "pods": 50
+  "RemovePodsHavingTooManyRestarts":
+     enabled: true
+     params:
+       podsHavingTooManyRestarts:
+         podeRestartThresholds: 100
+         includingInitContainers: true