openclaw - 💡(How to fix) Fix [Bug]: qqbot + ollama + local model: qwen3.5:27b report: error Embedded agent failed before reply: LLM request failed: network connection was interrupted

Bug type

Regression (worked before, now fails)

Beta release blocker

No

Summary

I report the bug as #85271. And I see it's fixed in release 2026.5.26-beta.1. But doing my local teat, still report LLM related warnings. Not sure any other block points or not:

18:32:32+08:00 warn agent/embedded {"subsystem":"agent/embedded"} {"event":"auth_profile_failure_state_updated","tags":["error_handling","auth_profiles","cooldown"],"runId":"843bf362-e243-4416-8178-6ad967b63c11","profileId":"sha256:9c018ec112cf","provider":"ollama","reason":"timeout","windowType":"cooldown","windowReused":false,"previousErrorCount":1,"errorCount":1,"previousCooldownUntil":1779877766437,"cooldownUntil":1779877982500,"failureCounts":{"timeout":1}} auth profile failure state updated 18:32:32+08:00 warn agent/embedded {"subsystem":"agent/embedded"} {"event":"embedded_run_failover_decision","tags":["error_handling","failover","assistant","surface_error"],"runId":"843bf362-e243-4416-8178-6ad967b63c11","stage":"assistant","decision":"surface_error","failoverReason":"timeout","profileFailureReason":"timeout","provider":"ollama","model":"qwen3.6:27b","sourceProvider":"ollama","sourceModel":"qwen3.6:27b","profileId":"sha256:9c018ec112cf","fallbackConfigured":false,"timedOut":false,"aborted":false,"rawErrorPreview":"terminated | read ECONNRESET","rawErrorHash":"sha256:ed1fcd7846a1","providerRuntimeFailureKind":"timeout"} embedded run failover decision 18:32:32+08:00 error diagnostic {"subsystem":"diagnostic"} lane task error: lane=main durationMs=106318 error="FailoverError: LLM request failed: network connection was interrupted." 18:32:32+08:00 error diagnostic {"subsystem":"diagnostic"} lane task error: lane=session:agent:main:qqbot:direct:d60c472a94703b2b231bec80dc343327 durationMs=106319 error="FailoverError: LLM request failed: network connection was interrupted." 18:32:32+08:00 warn model-fallback/decision {"subsystem":"model-fallback/decision"} {"event":"model_fallback_decision","tags":["error_handling","model_fallback","candidate_failed"],"runId":"843bf362-e243-4416-8178-6ad967b63c11","sessionId":"aef02403-c127-4a77-93c0-d630fb1a9a1e","lane":"main","decision":"candidate_failed","requestedProvider":"ollama","requestedModel":"qwen3.6:27b","candidateProvider":"ollama","candidateModel":"qwen3.6:27b","attempt":1,"total":1,"reason":"timeout","status":408,"errorPreview":"terminated | read ECONNRESET","errorHash":"sha256:ed1fcd7846a1","fallbackStepType":"fallback_step","fallbackStepFromModel":"ollama/qwen3.6:27b","fallbackStepFromFailureReason":"timeout","fallbackStepFromFailureDetail":"terminated | read ECONNRESET","fallbackStepChainPosition":1,"fallbackStepFinalOutcome":"chain_exhausted","isPrimary":true,"requestedModelMatched":true,"fallbackConfigured":false} model fallback decision 18:32:32+08:00 error Embedded agent failed before reply: LLM request failed: network connection was interrupted. 18:32:33+08:00 info gateway {"subsystem":"gateway"} provider auth state re-warmed (auth-profile-failure) in 10ms eventLoopMax=0.0ms 18:32:33+08:00 info channels/qqbot {"subsystem":"channels/qqbot"} [default] [qqbot:api] <<< Status: 200 OK | TraceId: 14af253b677934d7d5051f2f0574fbcd 18:32:33+08:00 info channels/qqbot {"subsystem":"channels/qqbot"} [default] onMessageSent called: refIdx=REFIDX_GG1J+2xJNRUoo0ve+6PYFA==, mediaType=undefined, ttsText=undefined 18:32:33+08:00 info channels/qqbot {"subsystem":"channels/qqbot"} [default] Sent markdown chunk (110/110 chars) with 0 HTTP images (c2c) v18:47:54+08:00 info channels/qqbot {"subsystem":"channels/qqbot"} [default] WebSocket closed: 4009 Session timed out 18:47:54+08:00 info channels/qqbot {"subsystem":"channels/qqbot"} [default] Error 4009 (session timed out), will re-identify 18:47:57+08:00 info channels/qqbot {"subsystem":"channels/qqbot"} [default] ✅ Access token obtained successfully 18:47:57+08:00 info channels/qqbot {"subsystem":"channels/qqbot"} [default] [qqbot:api] <<< Status: 200 OK | TraceId: fb8c373a4d4c16f9b9546a6c3290d0ef 18:47:57+08:00 info channels/qqbot {"subsystem":"channels/qqbot"} [default] Connecting to wss://api.sgroup.qq.com/websocket 18:47:57+08:00 info channels/qqbot {"subsystem":"channels/qqbot"} [default] WebSocket connected 18:47:57+08:00 info channels/qqbot {"subsystem":"channels/qqbot"} [default] [qqbot:token:1903883580] Background refresh already running 18:47:58+08:00 info channels/qqbot {"subsystem":"channels/qqbot"} [qqbot:default] Gateway ready

Steps to reproduce

When update to 2026.5.26-beta.1, warning changes to LLM request failed: network connection was interrupted.

Expected behavior

Should run as "timeoutSeconds": 1800 timeout without any warning ahead of time.

Actual behavior

qqbot just shows "Something ent wrong while processing your request. Please try again, or use /new to start a fresh session."

OpenClaw version

2026.5.26-beta.1

Operating system

centos os stream 8 with latest patches

Install method

No response

Model

2026.5.26-beta.1 + ollama + local model: qwen 3.5:27b or even use qwen 3.6:27b

Provider / routing chain

openclaw -> ollama -> qwen local model

Additional provider/model setup details

No response

Logs, screenshots, and evidence

Impact and severity

No response

Additional information

No response