renamed prefix-cache to prefix-cache-scorer

Signed-off-by: Nir Rozenbaum <[email protected]>

Author: nirrozenbaum

config/manifests/inferencepool-resources.yaml

@@ -152,7 +152,7 @@ data:
     plugins:
     - type: queue-scorer
     - type: kv-cache-scorer
-    - type: prefix-cache
+    - type: prefix-cache-scorer
       parameters:
         hashBlockSize: 64
         maxPrefixBlocksToMatch: 256
@@ -168,7 +168,7 @@ data:
         weight: 1
       - pluginRef: kv-cache-scorer
         weight: 1
-      - pluginRef: prefix-cache
+      - pluginRef: prefix-cache-scorer
         weight: 1
       - pluginRef: max-score-picker
 ---

pkg/epp/common/config/loader/configloader_test.go

@@ -657,8 +657,8 @@ plugins:
   type: low-queue-filter
   parameters:
     threshold: 10
-- name: prefixCache
-  type: prefix-cache
+- name: prefixCacheScorer
+  type: prefix-cache-scorer
   parameters:
     hashBlockSize: 32
 - name: maxScorePicker
@@ -669,7 +669,7 @@ schedulingProfiles:
 - name: default
   plugins:
   - pluginRef: lowQueueFilter
-  - pluginRef: prefixCache
+  - pluginRef: prefixCacheScorer
     weight: 50
   - pluginRef: maxScorePicker
 `
@@ -683,14 +683,14 @@ kind: EndpointPickerConfig
 plugins:
 - name:profileHandler
   type: single-profile
-- name: prefixCache
-  type: prefix-cache
+- name: prefixCacheScorer
+  type: prefix-cache-scorer
   parameters:
     hashBlockSize: asdf
 schedulingProfiles:
 - name: default
   plugins:
-  - pluginRef: prefixCache
+  - pluginRef: prefixCacheScorer
     weight: 50
 `
 
@@ -703,14 +703,14 @@ kind: EndpointPickerConfig
 plugins:
 - name: profileHandler
   type: single-profile
-- name: prefixCache
-  type: prefix-cache
+- name: prefixCacheScorer
+  type: prefix-cache-scorer
   parameters:
     hashBlockSize: 32
 schedulingProfiles:
 - name: default
   plugins:
-  - pluginRef: prefixCache
+  - pluginRef: prefixCacheScorer
 `
 
 // multiple pickers in scheduling profile

pkg/epp/scheduling/framework/plugins/multi/prefix/plugin.go

@@ -52,7 +52,7 @@ const (
 	// in vLLM, we will have 250K / 16 = 31.25K blocks.
 	DefaultLRUCapacityPerServer = 31250
 
-	PrefixCachePluginType = "prefix-cache"
+	PrefixCachePluginType = "prefix-cache-scorer"
 )
 
 type Config struct {

test/testdata/inferencepool-e2e.yaml

@@ -154,7 +154,7 @@ data:
     plugins:
     - type: queue-scorer
     - type: kv-cache-scorer
-    - type: prefix-cache
+    - type: prefix-cache-scorer
       parameters:
         hashBlockSize: 64
         maxPrefixBlocksToMatch: 256
@@ -170,7 +170,7 @@ data:
         weight: 1
       - pluginRef: kv-cache-scorer
         weight: 1
-      - pluginRef: prefix-cache
+      - pluginRef: prefix-cache-scorer
         weight: 1
       - pluginRef: max-score-picker
 ---