2026-02-03 11:28:22 +09:00
|
|
|
/**
|
|
|
|
|
* Runtime Fallback Hook - Constants
|
|
|
|
|
*
|
|
|
|
|
* Default values and configuration constants for the runtime fallback feature.
|
|
|
|
|
*/
|
|
|
|
|
|
|
|
|
|
import type { RuntimeFallbackConfig } from "../../config"
|
|
|
|
|
|
|
|
|
|
/**
|
|
|
|
|
* Default configuration values for runtime fallback
|
|
|
|
|
*/
|
|
|
|
|
export const DEFAULT_CONFIG: Required<RuntimeFallbackConfig> = {
|
2026-02-22 01:54:34 +09:00
|
|
|
enabled: false,
|
2026-04-06 17:38:37 +09:00
|
|
|
retry_on_errors: [429, 500, 502, 503, 504],
|
2026-02-03 11:28:22 +09:00
|
|
|
max_fallback_attempts: 3,
|
|
|
|
|
cooldown_seconds: 60,
|
2026-02-12 13:10:55 -05:00
|
|
|
timeout_seconds: 30,
|
2026-02-03 11:28:22 +09:00
|
|
|
notify_on_fallback: true,
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
/**
|
|
|
|
|
* Error patterns that indicate rate limiting or temporary failures
|
|
|
|
|
* These are checked in addition to HTTP status codes
|
|
|
|
|
*/
|
|
|
|
|
export const RETRYABLE_ERROR_PATTERNS = [
|
|
|
|
|
/rate.?limit/i,
|
|
|
|
|
/too.?many.?requests/i,
|
2026-03-04 18:35:09 +01:00
|
|
|
/quota\s+will\s+reset\s+after/i,
|
2026-04-06 17:38:37 +09:00
|
|
|
/quota.?exceeded/i,
|
2026-05-11 00:43:58 +08:00
|
|
|
/exceeded.*quota/i,
|
2026-05-11 01:03:02 +08:00
|
|
|
/usage\s*quota/i,
|
2026-03-04 18:35:09 +01:00
|
|
|
/exhausted\s+your\s+capacity/i,
|
2026-04-06 17:38:37 +09:00
|
|
|
/all\s+credentials\s+for\s+model/i,
|
|
|
|
|
/cool(?:ing)?\s+down/i,
|
2026-03-28 00:42:52 +09:00
|
|
|
/model.{0,20}?not.{0,10}?supported/i,
|
|
|
|
|
/model_not_supported/i,
|
2026-02-03 11:28:22 +09:00
|
|
|
/service.?unavailable/i,
|
|
|
|
|
/overloaded/i,
|
|
|
|
|
/temporarily.?unavailable/i,
|
|
|
|
|
/try.?again/i,
|
2026-02-05 23:15:05 +09:00
|
|
|
/(?:^|\s)429(?:\s|$)/,
|
|
|
|
|
/(?:^|\s)503(?:\s|$)/,
|
|
|
|
|
/(?:^|\s)529(?:\s|$)/,
|
2026-05-09 22:12:39 +08:00
|
|
|
|
|
|
|
|
// Chinese rate-limit / quota patterns (Zhipu, etc.)
|
|
|
|
|
/使用上限/, // "usage limit" — Zhipu: "已达到 5 小时的使用上限"
|
|
|
|
|
/频率限制/, // "rate limit" — generic Chinese rate-limit
|
|
|
|
|
/请求过于频繁/, // "too many requests" — common Chinese 429 message
|
|
|
|
|
/暂时不可用/, // "temporarily unavailable"
|
|
|
|
|
/服务不可用/, // "service unavailable"
|
|
|
|
|
/请稍后重试/, // "please try again later"
|
2026-02-03 11:28:22 +09:00
|
|
|
]
|
|
|
|
|
|
|
|
|
|
/**
|
|
|
|
|
* Hook name for identification and logging
|
|
|
|
|
*/
|
|
|
|
|
export const HOOK_NAME = "runtime-fallback"
|
2026-05-11 16:45:49 +01:00
|
|
|
|
|
|
|
|
/**
|
|
|
|
|
* First-prompt watchdog: how long to wait for the first sign of progress
|
|
|
|
|
* (assistant text/reasoning/finish) from a subagent session before assuming
|
|
|
|
|
* the provider is silently stuck and dispatching the configured fallback.
|
|
|
|
|
*
|
|
|
|
|
* Tuned to be longer than typical first-token latency (well under 30s in
|
|
|
|
|
* practice) yet much shorter than the 30-minute outer poll timeout that
|
|
|
|
|
* would otherwise be the only safety net.
|
|
|
|
|
*/
|
|
|
|
|
export const DEFAULT_FIRST_PROMPT_WATCHDOG_MS = 90_000
|
