9 Commits
Author SHA1 Message Date
oa-monitor 2e1dbb5403 fix: 14:41 updateBizState 卡死根因修复 + watchdog v8 优化
- ALTER ecology_message_info_read ADD INDEX idx_iread_target_biz (TARGETID, USERID, BIZSTATE)
- UPDATE ecology_message_info_read SET BIZSTATE=1 WHERE BIZSTATE=0 (清 164371+3437 条积压)
- watchdog v8: Layer 4 workflow_requestbase 联动检查 + Layer 5 BIZSTATE 积压 P1 告警
- 根因: DUMMY 补 397 万 info_read 后表膨胀 20x → updateBizState 单线程全表扫追不上 → BIZSTATE=0 积压 → message_log 停写
2026-11-13 16:57:20 +08:00
oa-monitor 875e466769 watchdog v7: HTTP 心跳前置 + Threads_running 辅助判定,解决 Deadlock 时 info_read/message_log 同时零增长盲区 2026-11-10 10:47:49 +08:00
oa-monitor 46ce657936 check-alert v3.1 + watchdog v6.1: 判定口径从 workflow_requestbase 改为 ecology_message_info_read 非 DUMMY,避免周末/刚提交就误报 2026-11-07 09:04:00 +08:00
oa-monitor cc3184fd3e check-alert v3: 改成 MySQL 流程驱动判定,和 watchdog v5 统一口径,周末不再误报推送断档 P0 2026-11-07 08:55:07 +08:00
oa-monitor 689303f5d7 watchdog v6: 自愈前先抓 jstack 现场,保存到 logs/jstack_dump/ 下次推送线程死锁根因可追溯 2026-11-06 18:28:51 +08:00
oa-monitor e8fab900fa watchdog v5: 流程驱动判定 - 有流程+无推送才自愈,避免误伤无流程时段 2026-11-06 16:01:28 +08:00
oa-monitor cb08965cf4 stop preventive restart cron, archive DUMMY_SYS script (根治完成) 2026-11-06 15:52:57 +08:00
oa-monitor 20fa48fb34 fix: check-alert v3 修复钉钉推送断档 P0 假阳性
根因:sqlite_exec 吞掉 stderr(2>/dev/null),查询失败返回空串,
0 默认为 0 导致误报

修复:业务层取数显式调用 sqlite3 + 检查退出码 + 正则校验,
查询失败跳过检测并 warn 日志
2026-11-04 19:41:18 +08:00
oa-monitor 7ab75364b1 feat: watchdog v4 + check-alert v2 + collect-business v2 + DUMMY补数据脚本 2026-11-04 17:25:42 +08:00