fix(网关): 修复账号选择中的调度器快照延迟问题
## 问题描述
调度器快照更新存在0.5-1秒的延迟(Outbox轮询间隔),导致在账号被限流或过载后的短时间窗口内,
可能仍会被选中,造成请求失败。
## 根本原因
账号选择逻辑依赖调度器快照(listSchedulableAccounts),但快照更新有延迟:
- Outbox轮询: 每1秒检查一次变更事件
- 全量重建: 每300秒重建一次
- 时间窗口: 账号状态变更后0.5-1秒内,快照可能未更新
## 解决方案
在账号选择循环中添加IsSchedulable()实时检查,作为第二道防线:
1. 第一道防线: 调度器快照过滤(可能有延迟)
2. 第二道防线: IsSchedulable()实时检查(本次修复)
IsSchedulable()会检查:
- RateLimitResetAt: 限流重置时间
- OverloadUntil: 过载持续时间
- TempUnschedulableUntil: 临时不可调度时间
- Status: 账号状态
- Schedulable: 可调度标志
## 修改范围
### OpenAI Gateway Service
- SelectAccountForModelWithExclusions: 添加IsSchedulable()检查
- SelectAccountWithLoadAwareness: 添加IsSchedulable()检查
### Gateway Service (Claude/Gemini/Antigravity)
- 负载感知选择候选账号筛选: 添加IsSchedulable()检查
- selectAccountForModelWithPlatform: 添加IsSchedulable()检查
- selectAccountWithMixedScheduling: 添加IsSchedulable()检查
### 测试用例
- OpenAI: 添加2个测试用例验证限流账号过滤
- Gateway: 添加2个测试用例验证限流和过载账号过滤
### 其他修复
- ops_repo_preagg.go: 修复platform为NULL时的聚合问题
## 测试结果
所有单元测试通过 ✅
This commit is contained in:
@@ -511,6 +511,12 @@ func (s *GatewayService) SelectAccountWithLoadAwareness(ctx context.Context, gro
|
||||
if isExcluded(acc.ID) {
|
||||
continue
|
||||
}
|
||||
// Scheduler snapshots can be temporarily stale (bucket rebuild is throttled);
|
||||
// re-check schedulability here so recently rate-limited/overloaded accounts
|
||||
// are not selected again before the bucket is rebuilt.
|
||||
if !acc.IsSchedulable() {
|
||||
continue
|
||||
}
|
||||
if !s.isAccountAllowedForPlatform(acc, platform, useMixed) {
|
||||
continue
|
||||
}
|
||||
@@ -893,6 +899,11 @@ func (s *GatewayService) selectAccountForModelWithPlatform(ctx context.Context,
|
||||
if _, excluded := excludedIDs[acc.ID]; excluded {
|
||||
continue
|
||||
}
|
||||
// Scheduler snapshots can be temporarily stale; re-check schedulability here to
|
||||
// avoid selecting accounts that were recently rate-limited/overloaded.
|
||||
if !acc.IsSchedulable() {
|
||||
continue
|
||||
}
|
||||
if !acc.IsSchedulableForModel(requestedModel) {
|
||||
continue
|
||||
}
|
||||
@@ -977,6 +988,11 @@ func (s *GatewayService) selectAccountWithMixedScheduling(ctx context.Context, g
|
||||
if _, excluded := excludedIDs[acc.ID]; excluded {
|
||||
continue
|
||||
}
|
||||
// Scheduler snapshots can be temporarily stale; re-check schedulability here to
|
||||
// avoid selecting accounts that were recently rate-limited/overloaded.
|
||||
if !acc.IsSchedulable() {
|
||||
continue
|
||||
}
|
||||
// 过滤:原生平台直接通过,antigravity 需要启用混合调度
|
||||
if acc.Platform == PlatformAntigravity && !acc.IsMixedSchedulingEnabled() {
|
||||
continue
|
||||
|
||||
Reference in New Issue
Block a user