k2safe/orbit
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

test: add reproducible schedule backlog rehearsal

Browse Source
This commit is contained in:
k2safe
2026-09-20 14:34:58 +08:00
parent 0d1026819c
commit e4225fd2ad
3 changed files with 144 additions and 2 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
docs/api/schedule-load.md
View File

@@ -20,7 +20,7 @@ go run ./cmd/orbit-schedule-load \
可以把 `-schedules` 调到 32、`-requests` 设为 4，观察同一秒集中触发及现有队列速率限制下的排队。队列默认并发和每秒授权量仍由部署配置决定，工具只记录预检快照，不提高配额。另可使用 `-schedules 2 -requests 12 -start-delay 1s -run-timeout 5s` 验证截止不足的失败报告；它必须非零退出，未观察到的触发点不能算成功。
要观察持久化积压，可在一次性数据库使用 `-schedules 32 -requests 8 -workers 1 -capacity 1 -start-delay 5s -run-timeout 180s`。该配置会让256个触发点集中进入队列，报告预期非零并保留未核验分母；独立连接应每2秒读取 `pulse_tasks` 的 `WAITING/LEASED/RUNNING/SUCCEEDED` 计数，示例结果见[长期积压报告](../reports/phase-6-backlog.md)。
要观察持久化积压，可在一次性数据库使用 `-schedules 32 -requests 8 -workers 1 -capacity 1 -start-delay 5s -run-timeout 180s`。该配置会让256个触发点集中进入队列，报告预期非零并保留未核验分母；独立连接应每2秒读取 `pulse_tasks` 的 `WAITING/LEASED/RUNNING/SUCCEEDED` 计数，示例结果见[长期积压报告](../reports/phase-6-backlog.md)。仓库还提供[`scripts/test-schedule-backlog.sh`](../../scripts/test-schedule-backlog.sh)自动创建/清理一次性 MySQL、采样状态和执行最终外键核验。
## 参数边界