Problemsby IbisvLLM runs out of memory
DevOps and infra2 posts from 2 people2d active+11 post in the last 7 days, 0 the 7 days before (steady)Unverified
Posts per day
The posts behind it
2, newest first[Bug]: Cold-cache KV cache OOM with `--kv-cache-memory-bytes` and `--enable-sleep-mode`### Your current environment RTX 5090 32 GB (SM120), driver 617.14, Windows 11 + WSL2 (Ubuntu 26.04), torch 2.13.0+cu130, FlashInfer 0.7.0.post1, vLLM nightly `0.32.0rc1.dev35+g656d0d450`. vllm collect-env ### 🐛 Describe the bug **Problem.** With `--kv-cache-memory-bytes` and...vllm-project/vllm · maxtonhuangOpen ↗[Bug]: EPLB async expert rearrangement crashes engine with AssertionError in fused MoE shared_experts (GLM-5.3-Flash, v0.31.0, TP8/EP8 + MTP)### Your current environment The output of python collect_env.py ### 🐛 Describe the bug `vllm serve zai-org/GLM-5.3-Flash \ --tensor-parallel-size 8 \ --enable-expert-parallel \ --enable-eplb \ --eplb-config.num-redundant-experts 8 \ --eplb-config.log-balancedness true \...vllm-project/vllm · yz3421 repliesOpen ↗
Companies and products named
About this problem
Evidence
2 posts from 2 people in 1 place, about 1 a week over 12 days. Mostly on vllm-project/vllm. Tools named alongside: NVIDIA.
Frustration Frustration 0 of 3· Seen on GitHub issues
How it was grouped
Posts that state a pain and match the "oom" rule about vLLM. First post Oct 7, 2026, latest Oct 18, 2026. Unverified: fewer than 3 posts, or one person or place only. Method
The same problem elsewhere
Gemini runs out of memoryJobs and services run out of memoryGitHub runs out of memoryNVIDIA runs out of memoryDuckDB runs out of memoryApache Iceberg runs out of memoryPolars runs out of memoryOllama runs out of memoryllama.cpp runs out of memoryApache Airflow runs out of memoryMistral runs out of memory
Other vLLM problems
vLLM upgrades break working setupsTrouble with vLLM decodingRunning vLLM locally hits memory and speed limitsvLLM GPU access and drivers are painfulRAG on vLLM returns irrelevant or missing contextTrouble with vLLM installationvLLM runs out of context on real workFine-tuning vLLM is costly and unpredictableTrouble with vLLM performancevLLM lock-in makes switching costlyvLLM users keep filing bugs
History
vLLM runs out of memory (2 posts)