fix precise-prefix-cache

Signed-off-by: threcc <trecchiu@redhat.com>

Author: threcc

tests/model_serving/model_server/llmd/llmd_configs/config_precise_prefix_cache.py

@@ -16,7 +16,7 @@ class PrecisePrefixCacheConfig(QwenHfConfig):
     model_name = "Qwen/Qwen2.5-7B-Instruct"
     replicas = 2
     block_size = 64
-    hash_algo = "sha256_cbor"
+    hash_algo = "sha256"
     hash_seed = "42"
     enable_auth = True
 
@@ -59,19 +59,16 @@ def _scheduler_config(cls):
                 {
                     "type": "precise-prefix-cache-scorer",
                     "parameters": {
+                        "tokenProcessorConfig": {
+                            "blockSize": cls.block_size,
+                            "hashSeed": cls.hash_seed,
+                        },
                         "kvEventsConfig": {"zmqEndpoint": "tcp://*:5557", "topicFilter": "kv"},
                         "indexerConfig": {
-                            "tokenProcessorConfig": {
-                                "blockSize": cls.block_size,
-                                "hashSeed": cls.hash_seed,
-                            },
                             "kvBlockIndexConfig": {
                                 "enableMetrics": True,
                                 "metricsLoggingInterval": 60000000000,
                             },
-                            "tokenizersPoolConfig": {
-                                "hf": {"tokenizersCacheDir": "/mnt/tokenizers"},
-                            },
                         },
                     },
                 },
@@ -101,8 +98,6 @@ def _scheduler_container(cls):
                 {"name": "metrics", "containerPort": 9090, "protocol": "TCP"},
                 {"name": "zmq", "containerPort": 5557, "protocol": "TCP"},
             ],
-            "env": [{"name": "HF_HOME", "value": "/mnt/tokenizers"}],
-            "volumeMounts": [{"name": "tokenizers", "mountPath": "/mnt/tokenizers", "readOnly": False}],
             "args": [
                 "--v=4",
                 "--pool-name",
@@ -133,7 +128,6 @@ def router_config(cls):
         return {
             "scheduler": {
                 "template": {
-                    "volumes": [{"name": "tokenizers", "emptyDir": {}}],
                     "containers": [cls._scheduler_container()],
                 }
             },

tests/model_serving/model_server/llmd/utils.py

@@ -342,10 +342,10 @@ def get_scheduler_decision_logs(
 
     # Get all logs from the scheduler pod
     # Note: The router-scheduler container is the default/main container
-    raw_logs = router_scheduler_pod.log()
+    raw_logs = router_scheduler_pod.log(container="main")
 
     # Target decision message
-    target_decision_msg = "Selecting pods from candidates sorted by max score"
+    target_decision_msg = "Selecting endpoints from candidates sorted by max score"
 
     # Filtering logs
     filtered_logs = "\n".join(line for line in raw_logs.splitlines() if target_decision_msg in line)