Simplify prometheus metrics (#1981)

Co-authored-by: Mohit Reddy <mohitreddy1996@users.noreply.github.com>
This commit is contained in:
Lianmin Zheng
2024-11-10 04:39:32 -08:00
committed by GitHub
parent ed53ac84b4
commit 1929c06762
11 changed files with 483 additions and 632 deletions

View File

@@ -17,7 +17,6 @@ limitations under the License.
import os
import random
import time
from collections import defaultdict
from contextlib import contextmanager
from enum import Enum, auto
@@ -307,7 +306,6 @@ class PrefillAdder:
):
# Non-chunked prefill
self.can_run_list.append(req)
req.queued_time = time.time()
self.tree_cache.inc_lock_ref(req.last_node)
self._prefill_one_req(
prefix_len,
@@ -326,7 +324,6 @@ class PrefillAdder:
req.extend_input_len = trunc_len
req.fill_ids = req.fill_ids[: len(req.prefix_indices) + trunc_len]
self.can_run_list.append(req)
req.queued_time = time.time()
self.new_inflight_req = req
self.tree_cache.inc_lock_ref(req.last_node)
self._prefill_one_req(prefix_len, trunc_len, 0)