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 14, 05:30 UTC
[misc] update the warning message (#25566)
Signed-off-by: youkaichao <youkaichao@gmail.com>
youkaichao committed
Sep 24, 2025 at 17:24 UTC
b67dece2d8a1fb8136948932a1203fdef6eb8981
1 file changed
+4
-2
vllm/distributed/device_communicators/shm_broadcast.py
+4
-2
@@ -392,7 +392,8 @@ class MessageQueue:
392
> VLLM_RINGBUFFER_WARNING_INTERVAL * n_warning):
393
logger.debug(
394
("No available shared memory broadcast block found"
395
- " in %s second."),
395
+ " in %s seconds. This typically happens when some"
396
+ " processes are hanging."),
397
VLLM_RINGBUFFER_WARNING_INTERVAL,
398
)
399
n_warning += 1
@@ -455,7 +456,8 @@ class MessageQueue:
456
> VLLM_RINGBUFFER_WARNING_INTERVAL * n_warning):
457
logger.debug(
458
("No available shared memory broadcast block found"
458
- " in %s second."),
459
+ " in %s seconds. This typically happens when some"
460
+ " processes are hanging."),
461
VLLM_RINGBUFFER_WARNING_INTERVAL,
462
)
463
n_warning += 1