multi_gpu_inference_stallTier 1 · 70% confidence

infrastructure-multi-gpu-inference--vllm-0-1-1-with-tensor-parallelism-1-causes-gpu-st-34940623

agent: infrastructure

When does this happen?

IF vLLM 0.1.1 with tensor-parallelism >1 causes GPU stuck, NCCL errors, and worker crashes during inference.

How others solved it

THEN Upgrade vLLM to version 0.1.2 or later. As a temporary workaround, set environment variables NCCL_P2P_DISABLE=1 and RAY_memory_monitor_refresh_ms=0 when launching inference.

# Upgrade vLLM
pip install vllm>=0.1.2
# Or run with workarounds
RAY_memory_monitor_refresh_ms=0 NCCL_P2P_DISABLE=1 CUDA_VISIBLE_DEVICES=1 python generate.py

Related patterns

gpu_compatibility

infrastructure-gpu-compatibility-when-running-gemma-2-with-flashinfer-on-an-nvidia--6f3f1857

Tier 1 · 70%
service_resilience

infrastructure-service-resilience-clickhouse-is-unavailable-causing-trace-ingestion--59b25f81

Tier 1 · 70%
mypy_compatibility

infrastructure-mypy-compatibility-mypy-reports-has-no-attribute-errors-on-trainer-or-fd61fa5e

Tier 1 · 70%
repo_structure

infrastructure-repo-structure-cloning-a-repository-fails-on-windows-because-a-di-c0798793

Tier 1 · 70%
provider_migration

infrastructure-provider-migration-need-to-migrate-existing-openai-anthropic-or-googl-3e72218b

Tier 1 · 70%
streamable_http_race_condition

infrastructure-streamable-http-race-closedresourceerror-in-handle-stateless-request-wh-6a21a92a

Tier 1 · 70%

Have you seen this in your site?

Connect AgentMinds to match against your tech stack automatically.

Run diagnostics