光合开发者社区-论坛

Z

zjf

楼主

4楼

我换qwen3.6+VLLM0.18+openclaw正常了，但是问1、2个问题后，qwen3.6疯狂重启了，报错：No available shared memory broadcast block found in 60 seconds. This typically happens when some processes are hanging or doing some time-consuming work (e.g., compilation, weight/kv cache quantization).这个是什么原因。

镜像：

harbor.sourcefind.cn:5443/dcu/admin/base/custom:vllm018-ubuntu22.04-dtk26.04-nemotron-20260422

4/30

L

ldwang

3楼

哪个镜像

4/29

Z

zjf

回复 @ldwang

我换qwen3.6+VLLM0.18+openclaw正常了，但是问1、2个问题后，qwen3.6疯狂重启了，报错：No available shared memory broadcast block found in 60 seconds. This typically happens when some processes are hanging or doing some time-consuming work (e.g., compilation, weight/kv cache quantization).这个是什么原因。

镜像：

harbor.sourcefind.cn:5443/dcu/admin/base/custom:vllm018-ubuntu22.04-dtk26.04-nemotron-20260422

4/30

Z

zjf

楼主

2楼

不带--enable-auto-tool-choice --tool-call-parser opena，报错：run error: 400 "auto" tool choice requires --enable-auto-tool-choice and --tool-call-parser to be set

4/29

Z

zjf

楼主

1楼

运行命令：vllm serve /data2/models/google/gemma-4-31B-it --port 8001 -tp 2 --trust-remote-code --host 0.0.0.0 --api-key "Jkzx@123" --served-model-name gemma-4 --enable-auto-tool-choice --tool-call-parser openai ；--tool-call-parser openai改为gemma4后报错：HTTP 400: ToolParser.init() takes 2 positional arguments but 3 were given

4/29