deprecate timeout

Signed-off-by: NickLucche <nlucches@redhat.com>

docker/Dockerfile

@@ -777,10 +777,8 @@ RUN --mount=type=cache,target=/root/.cache/uv \
                 libcublas-dev-${CUDA_VERSION_DASH} \
                 libcusolver-dev-${CUDA_VERSION_DASH}"; \
     if [ "$INSTALL_KV_CONNECTORS" = "true" ]; then \
-        # nixl metapackage ships with default nixl-cu12 (nixl<1.1.0). Explicitly strip nixl-cu13 
-        # from the requirements on non-cu13 images to prevent it from shadowing the correct cu12 one.
-        if [ "$CUDA_MAJOR" -lt 13 ]; then \
-            sed -i '/^nixl-cu13/d' /tmp/kv_connectors.txt; \
+        if [ "$CUDA_MAJOR" -ge 13 ]; then \
+            uv pip install --system nixl-cu13; \
         fi; \
         uv pip install --system -r /tmp/kv_connectors.txt --no-build || ( \
             # if the above fails, install from source

vllm/distributed/kv_transfer/kv_connector/v1/nixl/scheduler.py

@@ -2,6 +2,7 @@
 # SPDX-FileCopyrightText: Copyright contributors to the vLLM project
 """Scheduler-side logic for the NIXL connector."""
 
+import os
 import threading
 import time
 from typing import TYPE_CHECKING, Any
@@ -88,6 +89,12 @@ class NixlConnectorScheduler:
         if vllm_config.scheduler_config.disable_hybrid_kv_cache_manager:
             logger.info("Hybrid Memory Allocator is enabled with NIXL")
 
+        if os.environ.get("VLLM_NIXL_ABORT_REQUEST_TIMEOUT") is not None:
+            logger.warning(
+                "VLLM_NIXL_ABORT_REQUEST_TIMEOUT is deprecated and will be "
+                "removed in release 0.22.0."
+            )
+
         # Background thread for handling new handshake requests.
         self._nixl_handshake_listener_t: threading.Thread | None = None
         self._stop_event = threading.Event()