feat(server): hot-reload config store and reloadable infra clients

- Replace single Load() with a watching Store that re-reads config files
  (and config.local.yaml) and fans out a ReloadEvent with a per-field diff
  so consumers can decide whether the change is hot-applicable or requires
  a process restart.
- Wrap llm, retrieval, vector store, and object storage clients in
  Reloadable* shells so the bootstrap can swap their underlying impls when
  config changes without re-instantiating handlers.
- Make jwt.Manager and ops TokenIssuer mutable under a lock so secrets and
  TTLs can be rotated live; thread default plan code through a setter on
  the ops AdminUserService.
- Wire ConfigStore through bootstrap and every cmd/main.go, scheduler /
  worker / tenant-api / ops-api start the watcher; services and handlers
  take a config.Provider so they always read current values for things
  like generation.stream_enabled, scheduler dispatch, retrieval, etc.
- Switch shared/config decoding off viper to a Kratos-derived runtime
  package so env placeholders (\${VAR:default}) resolve consistently and
  the same source machinery powers both the loader and the watcher.

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
This commit is contained in:
2026-05-01 16:01:23 +08:00
parent ce2d8a2907
commit 618399f86d
61 changed files with 3186 additions and 496 deletions
@@ -13,6 +13,7 @@ import (
"go.uber.org/zap"
sharedcache "github.com/geo-platform/tenant-api/internal/shared/cache"
"github.com/geo-platform/tenant-api/internal/shared/config"
"github.com/geo-platform/tenant-api/internal/shared/llm"
"github.com/geo-platform/tenant-api/internal/shared/messaging/rabbitmq"
tenantapp "github.com/geo-platform/tenant-api/internal/tenant/app"
@@ -30,6 +31,7 @@ type KolAssistWorker struct {
cache sharedcache.Cache
retryInterval time.Duration
processTimeout time.Duration
configProvider config.Provider
consumerPrefix string
workerConcurrency int
}
@@ -65,6 +67,11 @@ func NewKolAssistWorker(
}
}
func (w *KolAssistWorker) WithConfigProvider(provider config.Provider) *KolAssistWorker {
w.configProvider = provider
return w
}
func (w *KolAssistWorker) Start(ctx context.Context) {
if w == nil || w.rabbitMQ == nil || w.repo == nil || w.llm == nil {
return
@@ -166,7 +173,7 @@ func (w *KolAssistWorker) handleDelivery(parent context.Context, delivery amqp.D
return
}
ctx, cancel := context.WithTimeout(parent, w.processTimeout)
ctx, cancel := context.WithTimeout(parent, w.runtimeProcessTimeout())
defer cancel()
if err := w.Process(ctx, *job); err != nil {
@@ -184,6 +191,25 @@ func (w *KolAssistWorker) handleDelivery(parent context.Context, delivery amqp.D
}
}
func (w *KolAssistWorker) runtimeProcessTimeout() time.Duration {
if w != nil && w.configProvider != nil {
if cfg := w.configProvider.Current(); cfg != nil {
return kolAssistProcessTimeout(cfg.Generation)
}
}
if w != nil && w.processTimeout > 0 {
return w.processTimeout
}
return kolAssistProcessTimeout(config.GenerationConfig{})
}
func kolAssistProcessTimeout(cfg config.GenerationConfig) time.Duration {
if cfg.ArticleTimeout > 0 {
return cfg.ArticleTimeout
}
return 5 * time.Minute
}
func (w *KolAssistWorker) rejectDelivery(delivery amqp.Delivery, requeue bool, err error, taskID string, tenantID int64) {
if nackErr := delivery.Nack(false, requeue); nackErr != nil && w.logger != nil {
w.logger.Warn("kol assist nack failed",