openclaw - 💡(How to fix) Fix [Bug]: Auth profile cooldown triggers chain exhaustion without actual Google API errors in v2026.5.26

Bug type

Regression (worked before, now fails)

Beta release blocker

No

Summary

On v2026.5.26, a single request failure causes OpenClaw to mark all Google auth profiles as cooldown simultaneously, exhausting the entire fallback chain (Flash → Lite → 2.5-Flash → Pro) within milliseconds. Google AI Studio confirms zero API errors at the same timestamp, proving requests never reached Google — the cooldown is triggered internally by OpenClaw before any real API call.

Steps to reproduce

Configure Google provider with multiple fallback models Observe occasional FailoverError in logs Check Google AI Studio usage — no corresponding 429 errors

Expected behavior

Cooldown should be scoped to the individual model that timed out, not the entire Google auth profile family.

Actual behavior

FailoverError: No available auth profile for google (all in cooldown or unavailable) All 4 fallback attempts fail within 186ms with reason=timeout, but Google AI Studio shows 0 API errors at the same timestamp.

User-visible symptom: Telegram displays: ⚠️ Something went wrong while processing your request. Please try again, or use /new to start a fresh session. Re-sending the same message immediately succeeds, confirming the issue is transient and self-recovering.

OpenClaw version

OpenClaw: 2026.5.26 (10ad3aa)

Operating system

Rocky Linux, Node 24.14.1

Install method

npm global

Model

google/gemini-3-flash-preview

Provider / routing chain

Telegram → OpenClaw Gateway → Google Generative AI (gemini-3-flash-preview with 3 fallbacks)

Additional provider/model setup details

"model": { "primary": "google/gemini-3-flash-preview", "fallbacks": [ "google/gemini-3.1-flash-lite", "google/gemini-2.5-flash", "google/gemini-3.1-pro-preview" ] }

Logs, screenshots, and evidence

{"subsystem":"diagnostic","message":"lane task error: lane=main durationMs=185 error=\"FailoverError: No available auth profile for google (all in cooldown or unavailable).\""}

{"subsystem":"model-fallback/decision","decision":"candidate_failed","requestedModel":"gemini-3-flash-preview","candidateModel":"gemini-3.1-pro-preview","attempt":4,"total":4,"reason":"timeout","fallbackStepFinalOutcome":"chain_exhausted"}

lane task error: lane=main durationMs=185 
error="FailoverError: No available auth profile for google (all in cooldown or unavailable)."

model fallback decision:
  attempt: 4, total: 4
  reason: timeout
  fallbackStepFinalOutcome: chain_exhausted
  candidateModel: gemini-3.1-pro-preview (last fallback)

Impact and severity

Severity: High — When triggered, the entire Google provider becomes unavailable for the affected session. Users receive ⚠️ Something went wrong with no response from the agent. Re-sending the message succeeds after the cooldown clears, but the failure is silent and confusing. Affected scope: Any configuration with multiple Google model fallbacks. All fallback candidates are contaminated simultaneously despite Google API being fully healthy (confirmed via Google AI Studio — zero 429 errors at the same timestamp).

Additional information

Last known good version: 2026.5.19. First known bad version: 2026.5.26. Workaround: re-sending the failed message succeeds immediately after the cooldown clears. Frequency: Intermittent — observed twice since upgrading to 2026.5.26 across multiple Telegram direct channels.