perf: improve worker and storage throughput

2026-07-26 00:57:29 +08:00
parent da9b273253
commit de5f451cd1
14 changed files with 611 additions and 101 deletions
--- a/docs/observability.md
+++ b/docs/observability.md
@@ -59,6 +59,8 @@ curl --fail http://127.0.0.1:18180/metrics
 - `imageforge_storage_fallbacks_total`
 - `imageforge_dead_letters_total`

+`imageforge_storage_fallbacks_total` 是存储容量风险信号，而不仅是普通降级统计。建议对 `increase(imageforge_storage_fallbacks_total[5m]) > 0` 持续 5 分钟设置高优先级告警，并同步监控应用服务器 `uploads` 卷使用率，避免 S3 长时间不可用时本地回退写满磁盘。
+
 Prometheus 抓取示例：

 ```yaml