flashduty · Oct 29, 2026
diff --git a/docs.json b/docs.json
@@ -3750,6 +3750,7 @@
                   "zh/rum/analytics/web",
                   "zh/rum/analytics/native",
                   "zh/rum/analytics/miniprogram",
+                  "zh/rum/analytics/chart-readings",
                   "zh/rum/analytics/custom-dashboards"
                 ]
               },
@@ -5546,6 +5547,7 @@
                   "en/rum/analytics/web",
                   "en/rum/analytics/native",
                   "en/rum/analytics/miniprogram",
+                  "en/rum/analytics/chart-readings",
                   "en/rum/analytics/custom-dashboards"
                 ]
               },

diff --git a/en/changelog/changelog.mdx b/en/changelog/changelog.mdx
@@ -4,6 +4,18 @@ description: "This page documents important updates and feature releases for Fla
 keywords: ["Changelog", "Product Release", "Feature Updates", "Flashduty", "Version History"]
 ---
 
+<Update label="2026-10-29" description="📉 RUM dashboard charts mark sampled estimates and can run one chart over all data, and the trend chart reads at 5 min / 1 hour / 1 day with a cumulative view">
+
+### RUM dashboards: estimate marks and exact queries per chart
+
+With a large amount of data, some dashboard charts are computed from a sample and carry an **Estimate** mark next to the title; hover over it to see roughly what fraction of the data was read. When you need the exact number, click **Exact query**: that chart alone is computed once over all data, the mark becomes **Exact**, and changing the time range or filters puts it back on estimates. An account may run a limited number of exact queries at once (2 by default); one that times out in the queue shows **Exact query failed**, and **Retry** runs it again. See [Chart readings](/en/rum/analytics/chart-readings).
+
+### RUM dashboards: granularity and cumulative view on the trend chart
+
+The **User Access Trend** chart on the overview page gains a **Granularity** dropdown (5 min, 1 hour, 1 day; 1 hour by default, and the dropdown itself says **Range too long** or **Range too short** when the current range cannot be drawn at that width) and a **Per interval / Cumulative** switch. Cumulative is the running total from the start of the selected range to that interval: UV counts each user once, so it is not the sum of per-interval UV, and you cannot drill down from a point; with 1 day, each bucket starts at midnight in your display time zone. See [Chart readings](/en/rum/analytics/chart-readings).
+
+</Update>
+
 <Update label="2026-10-25" description="🧭 Query and analyze in the RUM Explorer with AI SRE, a custom dashboard belongs to one application, errors split into Issues by where they were raised">
 
 ### RUM Explorer: query and analyze with AI SRE

diff --git a/en/monitors/quickstart/quickstart.mdx b/en/monitors/quickstart/quickstart.mdx
@@ -222,7 +222,7 @@ Event configuration runs from the channel through the notes description and repe
 | **Notes Description** | Unstructured text field; supports variable references to help responders quickly locate issues |
 | **Repeat Notification** | Continues notifying when alert is not recovered; configurable interval and maximum count (default 10000) |
 | **Custom Fields** | Similar to `annotations` in Prometheus; can attach dashboard URLs, SOP URLs, and so on. Click `+` next to the heading and fill in **Field name** and **Field value** in the dialog; the fields you add are listed as `key=value` tags — click a tag to edit it, click the × on it to delete it, and a field name cannot be empty or repeated |
-| **Related Query** | A separate subsection with its own add button; the result is not used for threshold evaluation, but can be referenced as a variable in the notes (for example, attach log samples) |
+| **Related Query** | A separate subsection with its own add button; the result is not used for threshold evaluation, but can be referenced as a variable in the notes (for example, attach log samples). When a related query fails to render, run, or read its fields on the alert engine, its result is left out of the `__alert_rule_relate_values` label and each failure is written to the `__alert_rule_relate_query_error` label as one "query name: error" line |
 
 <Note>
 The maximum notification count does not represent the number of message reminders end users receive. Because alert events generated by Monitors are delivered to On-call, they may be grouped and noise-reduced; the final send count depends on On-call configuration.

diff --git a/en/on-call/incident/search-view-incident.mdx b/en/on-call/incident/search-view-incident.mdx
@@ -163,6 +163,7 @@ The comprehensive info panel on the right side of the incident details page cons
 | **Responders** | Lists all responders (individuals or on-call personnel) assigned to the incident along with their acknowledgment status |
 | **Attributes** | Basic attributes such as channel, incident trigger time, and alert count |
 | **Investigation** | Appears only for incidents aggregated from Monitors alerts. Lists the investigation entries snapshot configured on the alert rule, showing 3 entries by default; below it you also get the alert rule that produced the incident with its main and related queries. See **Investigation** below |
+| **Issue info** | Appears only for incidents that come from RUM alerts (alerts carrying an `issue_id` label): the link text is that issue's error type and message, and clicking it opens the RUM issue in a new tab. When the issue cannot be loaded, the reason is shown in the section |
 | **Images** | Displayed when alerts associated with the incident include image information reported by API. Click a thumbnail to preview, and hover to view the Alt description and source link |
 | **Key Timestamps** | A timeline view of key timestamps in the incident lifecycle, including **trigger time**, **first acknowledgment time**, **close time**, and reopen events, so you can quickly grasp incident progress |
 | **Related Links** | Shortcut links preconfigured in the owning channel (such as dashboards, runbooks, knowledge bases). Parameters are dynamically rendered from the current incident's labels or fields, enabling one-click navigation to related systems. Link templates are maintained in channel settings |

diff --git a/en/rum/analytics/chart-readings.mdx b/en/rum/analytics/chart-readings.mdx
@@ -0,0 +1,36 @@
+---
+title: "Chart readings: estimates, exact queries, and time granularity"
+sidebarTitle: "Chart readings"
+description: "Read the Estimate mark on RUM dashboard charts, run one chart over all data, and switch the trend chart's granularity and cumulative view."
+keywords: ["RUM", "sampling", "estimate", "exact query", "granularity", "cumulative"]
+---
+
+## Estimates and exact queries
+
+With a large amount of data, not every chart on the analytics dashboards reads all of it: the **Estimate** mark next to the title means this chart was computed from a sample. Hover over the mark to see roughly what fraction of the data was read. Counts and sums are scaled up and ratios are computed on the sample, so the values are approximate.
+
+When you need the exact number for a chart, click **Exact query** next to the mark: that chart alone is computed once over all data, showing **Computing exactly…** while it runs and **Exact** when it is done. Changing the time range or the filters puts the chart back on sampled estimates.
+
+- An account may run a limited number of exact queries at once (2 by default) and the rest queue. A query that times out in the queue shows **Exact query failed**; click **Retry** to run it again.
+- A trend chart switched to **Cumulative** reads all data anyway, so it shows no estimate mark; if that chart would otherwise have been sampled, it uses the same allowance.
+
+## Trend chart granularity and cumulative view
+
+The **User Access Trend** chart on the overview page has two controls: the **Per interval / Cumulative** switch and the **Granularity** dropdown.
+
+| Control | Values | Description |
+|---------|--------|-------------|
+| **Trend view** | **Per interval** (default), **Cumulative** | Per interval is the value within each interval; cumulative is the running total from the start of the selected range to that interval |
+| **Granularity** | **5 min**, **1 hour** (default), **1 day** | The time each point covers; when the current range cannot be drawn at that width, an available one is used instead |
+
+- Every point in the cumulative view covers everything from the start of the range, so you cannot drill down from one. UV counts each user once, so the end of the cumulative line is not the sum of per-interval UV.
+- When a granularity cannot draw the current range, the dropdown says why: **Range too long** (too many points) or **Range too short** (fewer than two points).
+- With **1 day**, each bucket starts at midnight in your display time zone (the one in [personal settings](/en/on-call/configuration/personal-settings), or the [organization time zone](/en/platform/organization-info) when you have not set your own), not at midnight UTC.
+- With a large amount of data, the summary values on the left of the card are sampled estimates and may differ slightly from the end of the cumulative line; after **Exact query** on that chart the two agree.
+
+## Related pages
+
+- [Web RUM analytics](/en/rum/analytics/web)
+- [Native](/en/rum/analytics/native)
+- [WeChat Mini Program](/en/rum/analytics/miniprogram)
+- [RUM custom dashboards](/en/rum/analytics/custom-dashboards)
diff --git a/zh/changelog/changelog.mdx b/zh/changelog/changelog.mdx
@@ -4,6 +4,18 @@ description: "本页面记录 Flashduty 产品的重要更新和功能发布"
 keywords: ["更新日志", "产品发布", "功能更新", "Flashduty", "版本记录"]
 ---
 
+<Update label="2026-10-29" description="📉 RUM 看板的图表会标记抽样估算、可对单张图跑一次全量计算，趋势图能按 5 分钟 / 1 小时 / 1 天查看并切换累积">
+
+### RUM 看板：估算标记与单图精确查询
+
+数据量大时，分析看板的部分图表按样本计算，标题旁显示 **估算** 标记，悬停可以看到本次统计了约几分之一的数据。需要准确数字时点 **精确查询**：只对这张图读一次全量数据，完成后标记变成 **精确**，改时间范围或筛选条件后回到估算。同一账户同时运行的精确查询默认最多 2 条，排队超时会显示 **精确查询失败**，点 **重试** 再跑一次。详见 [图表读数](/zh/rum/analytics/chart-readings)。
+
+### RUM 看板：趋势图的聚合粒度与累积视图
+
+概览页的 **用户访问趋势** 图新增 **聚合粒度** 下拉（5 分钟、1 小时、1 天，默认 1 小时；当前时间范围画不出来时下拉里直接注明 **时间范围过长** 或 **时间范围过短**）与 **分时 / 累积** 切换。累积是从所选时间起点累加到该时间段的量：UV 按用户去重，不等于各时段 UV 相加，且不能点击下钻；选 1 天时每个桶从您展示时区的零点开始。详见 [图表读数](/zh/rum/analytics/chart-readings)。
+
+</Update>
+
 <Update label="2026-10-25" description="🧭 RUM 查看器可以用 AI SRE 查询和分析、自定义看板归属到单个应用、异常按出错位置拆成不同 Issue">
 
 ### RUM 查看器：用 AI SRE 查询和分析

diff --git a/zh/monitors/quickstart/quickstart.mdx b/zh/monitors/quickstart/quickstart.mdx
@@ -223,7 +223,7 @@ Go 模板可以使用以下变量：
 | **备注描述** | 非结构化文本字段，支持引用变量，便于值班人员快速定位问题 |
 | **重复通知** | 告警未恢复时持续通知，可配置间隔和最大次数（默认 10000 次） |
 | **自定义字段** | 类似 Prometheus 中的 `annotations`，可附加仪表盘 URL、SOP URL 等。点标题右侧的 `+` 在弹窗里填 **字段名** 与 **字段值**；已添加的字段以 `键=值` 标签列出，点标签可再次修改、点标签上的 × 删除，字段名不能为空也不能重复 |
-| **关联查询** | 独立小节，有自己的新增按钮；查询结果不作为阈值判定依据，但可在备注中作为变量引用（如附加日志样例） |
+| **关联查询** | 独立小节，有自己的新增按钮；查询结果不作为阈值判定依据，但可在备注中作为变量引用（如附加日志样例）。某条关联查询在告警引擎上渲染、执行或取字段失败时，它的结果不出现在 `__alert_rule_relate_values` 标签里，失败原因按「查询名: 错误」逐条写入 `__alert_rule_relate_query_error` 标签 |
 
 <Note>
 最大通知次数并不代表终端用户收到的消息提醒次数。因为 Monitors 产生的告警事件会投递到 On-call，可能会被聚合降噪，最终发送次数取决于 On-call 配置。

diff --git a/zh/on-call/incident/search-view-incident.mdx b/zh/on-call/incident/search-view-incident.mdx
@@ -164,6 +164,7 @@ AI 总结仅适用于由告警自动触发的故障，手动创建的故障不
 | **处理人员** | 列出分派到该故障的所有处理人员（个人或值班人员）及其认领状态 |
 | **属性** | 协作空间、故障触发时间、告警数量等基础属性 |
 | **排查入口** | 仅当故障由 Monit 告警聚合而成时出现，列出告警规则里配置的排查入口快照，默认展示 3 条；下方还会给出产生该故障的告警规则与其主查询 / 关联查询。详见下方 **排查入口** |
+| **Issue 信息** | 仅当故障来自 RUM 告警（告警带 `issue_id` 标签）时出现：链接文字是该 Issue 的错误类型与错误消息，点击在新页签打开对应的 RUM Issue。读取失败时区块内显示失败原因 |
 | **图片** | 当故障关联的告警通过 API 上报了图片信息时展示。点击缩略图可预览，悬浮可查看图片的 Alt 描述和来源链接 |
 | **关键时间节点** | 以时间轴形式展示故障生命周期中的关键时间戳，包括 **触发时间**、**首次认领时间**、**关闭时间** 以及重新打开等节点，便于快速掌握故障进展 |
 | **关联链接** | 所属协作空间预配置的快捷链接（如监控大盘、Runbook、知识库），支持基于当前故障的标签或字段动态渲染参数，便于一键跳转到相关系统。关联链接的模板在协作空间设置中维护 |

diff --git a/zh/rum/analytics/chart-readings.mdx b/zh/rum/analytics/chart-readings.mdx
@@ -0,0 +1,36 @@
+---
+title: "图表读数：估算、精确查询与时间粒度"
+sidebarTitle: "图表读数"
+description: "看懂 RUM 分析看板图表上的「估算」标记，对单张图跑一次全量计算，并切换趋势图的聚合粒度与累积视图。"
+keywords: ["RUM", "抽样", "估算", "精确查询", "聚合粒度", "累积"]
+---
+
+## 估算与精确查询
+
+数据量很大时，分析看板上的图表不一定读全量数据，标题旁的 **估算** 标记表示这张图按样本计算。悬停标记可以看到本次统计了约几分之一的数据：计数与求和按比例放大、比率按样本计算，所以数值是近似值。
+
+需要这张图的准确数字时，点标记旁的 **精确查询**：只对这张图读一次全量数据，期间显示 **精确计算中…**，完成后标记变成 **精确**。改动时间范围或筛选条件后，这张图会回到抽样估算。
+
+- 同一账户同时运行的精确查询数量有限（默认 2 条），超出的排队等待。排队等到超时会显示 **精确查询失败**，点 **重试** 再跑一次。
+- 趋势图切到 **累积** 时本来就要读全量数据，所以不显示估算标记；这张图如果本来会被抽样，它同样占用上面这条额度。
+
+## 趋势图的聚合粒度与累积视图
+
+概览页的 **用户访问趋势** 图提供两个控件：**分时 / 累积** 切换，以及 **聚合粒度** 下拉。
+
+| 控件 | 可选值 | 说明 |
+|------|--------|------|
+| **趋势视图** | **分时**（默认）、**累积** | 分时是每个时间段内的数值；累积是从所选时间起点累加到该时间段的总量 |
+| **聚合粒度** | **5 分钟**、**1 小时**（默认）、**1 天** | 一个点覆盖的时间宽度；当前时间范围下不适用时，自动改用可用的粒度 |
+
+- 累积视图里每个点都代表从起点到该点的全部数据，所以不能点击下钻。UV 按用户去重，累积的终点不是各时段 UV 相加。
+- 粒度在当前时间范围下画不出来时，下拉里直接给出原因：**时间范围过长**（点数超过上限）或 **时间范围过短**（不足两个点）。
+- 选 **1 天** 时，每个桶从您展示时区的零点开始（[个人设置](/zh/on-call/configuration/personal-settings)里的时区，未单独设置时跟随[组织时区](/zh/platform/organization-info)），不是 UTC 零点。
+- 数据量大时，卡片左侧的汇总值是抽样估算，可能与累积的终点略有出入；对这张图点 **精确查询** 后两者一致。
+
+## 相关页面
+
+- [Web RUM 分析](/zh/rum/analytics/web)
+- [Native](/zh/rum/analytics/native)
+- [微信小程序](/zh/rum/analytics/miniprogram)
+- [RUM 自定义看板](/zh/rum/analytics/custom-dashboards)