Use the correct vllm metric gpu_cache_usage_perc --> kv_cache_usage_perc (#1905)

Signed-off-by: Ezra Silvera <ezra@il.ibm.com>

Author: ezrasilvera

docs/proposals/003-model-server-protocol/README.md

@@ -29,7 +29,7 @@ effort.
 | ----- | ---- | ------------ | ---- | ---- | ---- |
 | TotalQueuedRequests         | Gauge     | The current total number of requests in the queue.| `vllm:num_requests_waiting`| `nv_trt_llm_request_metrics{request_type=waiting}`| `sglang:num_queue_reqs`
 | TotalRunningRequests         | Gauge     | The current total number of requests actively being served on the model server.| `vllm:num_requests_running`| `nv_trt_llm_request_metrics{request_type=scheduled}`| `sglang:num_running_reqs`
-| KVCacheUtilization| Gauge     | The current KV cache utilization in percentage.| `vllm:gpu_cache_usage_perc`| `nv_trt_llm_kv_cache_block_metrics{kv_cache_block_type=fraction}`| `sglang:token_usage`
+| KVCacheUtilization| Gauge     | The current KV cache utilization in percentage.| `vllm:kv_cache_usage_perc`| `nv_trt_llm_kv_cache_block_metrics{kv_cache_block_type=fraction}`| `sglang:token_usage`
 | [Optional] BlockSize         | Labeled     | The block size in tokens to allocate memory, used by the prefix cache scorer. If this metric is not available, the BlockSize will be derived from the [prefix plugin config](https://gateway-api-inference-extension.sigs.k8s.io/guides/epp-configuration/prefix-aware/#customize-the-prefix-cache-plugin).| name: `vllm:cache_config_info`, label name: `block_size`| | 
 | [Optional] NumGPUBlocks| Labeled     | The total number of blocks in the HBM KV cache, used by the prefix cache scorer. If this metric is not available, the NumGPUBlocks will be derived from the [prefix plugin config](https://gateway-api-inference-extension.sigs.k8s.io/guides/epp-configuration/prefix-aware/#customize-the-prefix-cache-plugin).| name: `vllm:cache_config_info`, label name: `num_gpu_blocks`| | 
 

pkg/epp/datalayer/metrics/extractor_test.go

@@ -32,7 +32,7 @@ const (
 	// use hardcoded values - importing causes cycle
 	defaultTotalQueuedRequestsMetric    = "vllm:num_requests_waiting"
 	defaultTotalRunningRequestsMetric   = "vllm:num_requests_running"
-	defaultKvCacheUsagePercentageMetric = "vllm:gpu_cache_usage_perc"
+	defaultKvCacheUsagePercentageMetric = "vllm:kv_cache_usage_perc"
 	defaultLoraInfoMetric               = "vllm:lora_requests_info"
 	defaultCacheInfoMetric              = "vllm:cache_config_info"
 )

pkg/epp/server/runserver.go

@@ -79,7 +79,7 @@ const (
 	DefaultEnablePprof                      = true                          // default for --enable-pprof
 	DefaultTotalQueuedRequestsMetric        = "vllm:num_requests_waiting"   // default for --total-queued-requests-metric
 	DefaultTotalRunningRequestsMetric       = "vllm:num_requests_running"   // default for --total-running-requests-metric
-	DefaultKvCacheUsagePercentageMetric     = "vllm:gpu_cache_usage_perc"   // default for --kv-cache-usage-percentage-metric
+	DefaultKvCacheUsagePercentageMetric     = "vllm:kv_cache_usage_perc"    // default for --kv-cache-usage-percentage-metric
 	DefaultLoraInfoMetric                   = "vllm:lora_requests_info"     // default for --lora-info-metric
 	DefaultCacheInfoMetric                  = "vllm:cache_config_info"      // default for --cache-info-metric
 	DefaultCertPath                         = ""                            // default for --cert-path