ikmkj/new-api
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Align console behavior with operator routing and image workflow expectations

Browse Source 
Constraint: Must preserve current exchange-rate logic, same-group routing boundary, and temporary image-file behavior while fixing UX-visible inconsistencies.
Rejected: Large frontend chunk refactor | introduced risky circular chunking and did not improve deliverability for this rollout.
Confidence: medium
Scope-risk: moderate
Directive: Keep channel failover constrained to the selected group and preserve Beijing-day semantics for user cache dashboard metrics.
Tested: go test ./controller ./model ./service ./common -count=1; npm run build; remote smoke on source deployment at http://38.76.218.56:18081/ plus /api/status and /api/image-studio/settings.
Not-tested: Browser-driven end-to-end verification of every console page interaction.
This commit is contained in:
ikmkj
2026-05-05 15:27:51 +08:00
parent 92ee465a41
commit 8ee89478a9
46 changed files with 1341 additions and 289 deletions
Download Patch File Download Diff File Expand all files Collapse all files

16
service/channel_select.go
View File

@@ -45,14 +45,14 @@ func (p *RetryParam) ResetRetryNextTry() {
p.resetNextTry = true
}
// CacheGetRandomSatisfiedChannel tries to get a random channel that satisfies the requirements.
// 尝试获取一个满足要求的随机渠道。
// CacheGetRandomSatisfiedChannel tries to get the next satisfied channel within the same group.
// 按同一分组内的优先级和顺序选择下一个满足要求的渠道。
//
// For "auto" tokenGroup with cross-group Retry enabled:
// 对于启用了跨分组重试的 "auto" tokenGroup：
//
// - Each group will exhaust all its priorities before moving to the next group.
// 每个分组会用完所有优先级后才会切换到下一个分组。
// - Each group will exhaust all its channels before moving to the next group.
// 每个分组会用完所有候选渠道后才会切换到下一个分组。
//
// - Uses ContextKeyAutoGroupIndex to track current group index.
// 使用 ContextKeyAutoGroupIndex 跟踪当前分组索引。
@@ -60,11 +60,11 @@ func (p *RetryParam) ResetRetryNextTry() {
// - Uses ContextKeyAutoGroupRetryIndex to track the global Retry count when current group started.
// 使用 ContextKeyAutoGroupRetryIndex 跟踪当前分组开始时的全局重试次数。
//
// - priorityRetry = Retry - startRetryIndex, represents the priority level within current group.
// priorityRetry = Retry - startRetryIndex，表示当前分组内的优先级级别。
// - priorityRetry = Retry - startRetryIndex, represents the channel attempt index within current group.
// priorityRetry = Retry - startRetryIndex，表示当前分组内的候选渠道尝试序号。
//
// - When GetRandomSatisfiedChannel returns nil (priorities exhausted), moves to next group.
// 当 GetRandomSatisfiedChannel 返回 nil（优先级用完）时，切换到下一个分组。
// - When GetRandomSatisfiedChannel returns nil (group exhausted), moves to next group.
// 当 GetRandomSatisfiedChannel 返回 nil（当前分组候选已用尽）时，切换到下一个分组。
//
// Example flow (2 groups, each with 2 priorities, RetryTimes=3):
// 示例流程（2个分组，每个有2个优先级，RetryTimes=3）：