【运维发布】故障复盘：事故报告与改进跟踪

---事故报告的模板结构
local REPORT_TMPL = {
    title = "",        -- 一句话事故名
    level = "",        -- P0/P1/P2
    timeline = {},     -- 时间轴：发生/发现/响应/恢复
    impact = {},       -- 影响：范围/时长/损失
    rootCause = "",    -- 根因：技术链路的分析
    why = {},          -- 五问：逐层追问
    actions = {},      -- 改进项：措施/负责人/期限
}

function Postmortem.create(incidentId)
    local rpt = {}
    for k, v in pairs(REPORT_TMPL) do
        rpt[k] = type(v) == "table" and {} or v
    end
    rpt.incidentId = incidentId
    Postmortem.store[incidentId] = rpt
    return rpt
end

---改进项的跟踪：到期未完成的升级
setontimerex(98, 3600)
function Postmortem.track()
    local now = os.time()
    for _, rpt in pairs(Postmortem.store) do
        for _, act in ipairs(rpt.actions or {}) do
            if not act.done and now > act.deadline then
                Ops.alert("改进项逾期", rpt.title,
                    act.task, act.owner)
                act.escalated = true
            end
        end
    end
end