[kv_offload] fix(metrics): rename kv_offload_tiering_block_{queries,hits} → chunk (#52812)
Signed-off-by: Ronen Schaffer <ronen.schaffer@ibm.com>
Ronen Schaffer committed
Aug 20, 2026 at 03:24 UTC
0a111cca2cddbfccf5b99f61af9c9bbde53f6d9c
1 file changed
+2
-2
vllm/v1/kv_offload/tiering/base.py
+2
-2
@@ -45,8 +45,8 @@ class TieringOffloadingMetrics:
45
WRITE_TIME = "vllm:kv_offload_tiering_write_time"
46
PROMOTION_JOB_FAILURES = "vllm:kv_offload_tiering_promotion_job_failures"
47
CASCADE_JOB_FAILURES = "vllm:kv_offload_tiering_cascade_job_failures"
48
- BLOCK_QUERIES = "vllm:kv_offload_tiering_block_queries"
49
- BLOCK_HITS = "vllm:kv_offload_tiering_block_hits"
48
+ BLOCK_QUERIES = "vllm:kv_offload_tiering_chunk_queries"
49
+ BLOCK_HITS = "vllm:kv_offload_tiering_chunk_hits"
50
PRIMARY_WRITE_USAGE_PERC = "vllm:kv_offload_tiering_primary_write_usage_perc"
51
PRIMARY_READ_USAGE_PERC = "vllm:kv_offload_tiering_primary_read_usage_perc"
52
PROMOTION_ALLOCATION_FAILURES = (