OpenCode's Ollama discovery drops user timeout config, cutting long local generations at ~5 min
GigAnon · ghdev · 2026-10-13
- OpenCode's built-in Ollama discovery plugin takes over any provider named ollama, replacing the provider's package and settings — so user-configured timeout, headerTimeout and chunkTimeout are silently discarded in favor of 300,000 ms defaults.
- Result: any response streaming nothing for 5 minutes (e.g. a large tool call whose arguments are emitted only once complete, at 15 tok/s) is cut with Connection lost while reading the response.
- OpenCode retries up to 3 times, each failing the same way, while the abandoned request keeps the local model busy; pending tool calls report Tool execution interrupted.
- The issue was drafted by Claude during an automated debugging session, reviewed and filed by a human, based on code at v2.0.26 plus a real reproduction.
- Workaround: avoid naming the provider ollama so the discovery plugin doesn't take over.
More from coding & agent
- Ethan Mollick's new book charts 30 months of AI progress, from cyborg to conjurer work — rseroter · 2026-10-23
- OpenJet claims Qwen3.8-27B coding runs 1.9x faster than llama.cpp on one RTX 3090 — JeremyCMorgan · 2026-10-23
- Jensen Huang details Vera Rubin agentic stack: NVL72, MGX, CMX and RTX — rohanpaul_ai · 2026-10-23
- Building an autonomous SEO agent in Hermes with three skills: programmatic pages, Keywords Everywhere, DataForSEO — EXM7777 · 2026-10-23
- Building a sports betting AI agent with onchain execution and human approval by default: architecture and lessons — Dry-Comfortable-2678 · 2026-10-23
- Capy 0.5.0 lets Claude Pro/Max subscribers reuse their plan quota, adds OIDC cloud access — garrytan · 2026-10-23