Problemsby IbisRAG on vLLM returns irrelevant or missing context
LLM apps3 posts from 3 people3d active+11 post in the last 7 days, 1 the 7 days before (steady)
Posts per day
The posts behind it
3, newest first[Bug]: Multimodal requests stall when the long-prefill cap cannot fit an intact itemCorresponding PR: https://github.com/vllm-project/vllm/pull/62030 (created before this issue). ### Your current environment - Reproduced on `ef38e8d006c9835b351e7999ad28193efd81abdb` and rerun on current base `642c18d64c4cb6db3ded8fbe1786718cd4005fcb`, which underlies the...vllm-project/vllm · Justee21Open ↗[RFC]: TPSP Fusion Design### Motivation. # RFC: TPSP Fusion Design ## 0. Goal TPSP decomposes AllReduce into ReduceScatter (RS) + AllGather (AG), fusing RS/AG with adjacent compute (Norm, routing, GEMM) to hide communication behind computation. This brings roughly **10% TTFT improvement on long...vllm-project/vllm · hlin991 reactions, 2 repliesOpen ↗[Feature]: Production evidence for restoring concurrent partial prefill limits (max_num_partial_prefills / max_concurrent_prefills) in V1 scheduler...and we believe this is a strong real-world case for reviving #49075 / #55256. ## Problem With `--enable-chunked-prefill` and `--max-num-batched-tokens` set to a reasonably high value (tested 8192 and 16384), a single long-prompt request still needs many scheduler iterations to...vllm-project/vllm · javimp20031 reactions, 12 repliesOpen ↗
Companies and products named
About this problem
Evidence
3 posts from 3 people in 1 place, about 1 a week over 30 days. Mostly on vllm-project/vllm. Tools named alongside: NVIDIA, GitHub, Grafana, CodeRabbit.
Frustration Frustration 0 of 3· Seen on GitHub issues
How it was grouped
Posts that state a pain and match the "rag-retrieval" rule about vLLM. First post Sep 17, 2026, latest Oct 16, 2026. Corroborated: 3 or more posts from 2 or more people or places. Method
The same problem elsewhere
RAG retrieval returns irrelevant or missing contextRAG on GitHub returns irrelevant or missing contextRAG on LangChain returns irrelevant or missing contextRAG on Anthropic returns irrelevant or missing contextRAG on Claude returns irrelevant or missing contextRAG on PyTorch returns irrelevant or missing contextRAG on Open WebUI returns irrelevant or missing contextRAG on PostgreSQL returns irrelevant or missing contextRAG on Ollama returns irrelevant or missing contextRAG on Claude Code returns irrelevant or missing contextRAG on SQL Server returns irrelevant or missing contextRAG on LlamaIndex returns irrelevant or missing context
Other vLLM problems
vLLM upgrades break working setupsTrouble with vLLM decodingRunning vLLM locally hits memory and speed limitsvLLM GPU access and drivers are painfulTrouble with vLLM installationvLLM runs out of context on real workFine-tuning vLLM is costly and unpredictableTrouble with vLLM performancevLLM runs out of memoryvLLM lock-in makes switching costlyvLLM users keep filing bugs
History
unverified to corroborated
fading to steady
steady to fading
RAG on vLLM returns irrelevant or missing context (2 posts)