A high-throughput and memory-efficient inference and serving engine for LLMs
Mirrored from https://github.com/tduncan-lum/Nvidia-Optimization-Stack---VLLM.git
Synced Nov 15, 06:00 UTC
Add clear documentation around the impact of debugging flag (#19369)
Signed-off-by: Anna Pendleton <pendleton@google.com>
Anna Pendleton committed
Jun 9, 2025 at 23:16 UTC
32b3946bb4fb8c338ae6d1c04b7b1ef50896db55
1 file changed
+1
-1
docs/usage/troubleshooting.md
+1
-1
@@ -40,7 +40,7 @@ If other strategies don't solve the problem, it's likely that the vLLM instance
40
- `export VLLM_LOGGING_LEVEL=DEBUG` to turn on more logging.
41
- `export CUDA_LAUNCH_BLOCKING=1` to identify which CUDA kernel is causing the problem.
42
- `export NCCL_DEBUG=TRACE` to turn on more logging for NCCL.
43
-- `export VLLM_TRACE_FUNCTION=1` to record all function calls for inspection in the log files to tell which function crashes or hangs.
43
+- `export VLLM_TRACE_FUNCTION=1` to record all function calls for inspection in the log files to tell which function crashes or hangs. Do not use this flag unless absolutely needed for debugging, it will cause significant delays in startup time.
44
45
## Incorrect network setup
46