Merge pull request #733 from flashcatcloud/feat/alert-batch11-test3

docs(signoz): correct channel path, Test button behavior and severity source

Author: ysyneu

en/on-call/integration/alert-integration/alert-sources/signoz.mdx

@@ -32,10 +32,10 @@ Get an integration push URL in either of the two ways below. **Choose the Promet
 ## Configure in SigNoz
 ---
 
-1. In SigNoz, go to **Settings** → **Account Settings** → **Notification Channels** and click **New Channel**
+1. In SigNoz, go to **Settings** → **Notification Channels** (also reachable from the **Notification Channels** tab on the **Alerts** page) and click **New Channel**
 2. Enter a channel name, select **Webhook** as the type, and enter the full push URL copied above as the endpoint URL, including `?integration_key=...`
 3. Keep **Send resolved alerts** on (it is on by default); otherwise Flashduty receives no recovery notifications
-4. Save, then select the channel in the notification settings of your SigNoz alert rules
+4. Save, then select the channel in the threshold row (**send to**) of your SigNoz alert rules
 
 ## Field mapping
 ---
@@ -50,18 +50,19 @@ Get an integration push URL in either of the two ways below. **Choose the Promet
 | Other `labels` and `annotations` | Each key becomes a label; `annotations.description` becomes the alert description |
 | `startsAt`, `endsAt` | Labels `starts_at` and `ends_at` |
 
-SigNoz groups alerts by alert name and sends a grouped notification every 5 minutes; one notification can carry several alerts, and Flashduty handles each one separately (up to 100 per request).
+SigNoz groups notifications by rule (`ruleId`); one notification can carry several alerts, and Flashduty handles each one separately (up to 100 per request). A firing rule normally arrives within a minute.
 
 ## Recovery and deduplication
 ---
 
 - When SigNoz sends a notification with `status` set to `resolved`, Flashduty closes the alert with the same `fingerprint`.
-- The severity in SigNoz is a `severity` label on the rule, and its values are yours to define. If you use a value that is not built in, such as `error`, configure a custom severity mapping on the Prometheus integration; otherwise it is treated as Warning.
-- SigNoz's documentation does not describe what the Test button sends. After the first setup, verify with a real alert that both the trigger and the recovery show up in Flashduty.
+- In current SigNoz rules the `severity` label is taken from the threshold row name (for example a threshold named `critical` or `warning`), which is also written to the `threshold.name` label. If you use a value that is not built in, such as `error`, configure a custom severity mapping on the Prometheus integration; otherwise it is treated as Warning.
+- The resolved notification arrives several minutes after the rule returns to normal (observed: about 30 seconds for firing, 6 to 10 minutes for resolved). This is expected.
+- The channel's **Test** button sends a `firing` test alert titled `Test Alert (<channel name>)` with `severity` `critical`. It opens a Critical alert in Flashduty and is never followed by a resolved notification, so close that alert manually.
 
 ## Troubleshooting
 ---
 
 - **Flashduty returns `Invalid parameters`**: the push URL is incomplete and lacks `integration_key`, or the integration type is not Prometheus
 - **Alerts do not recover**: confirm **Send resolved alerts** is on for the SigNoz channel and that the resolved notification's `fingerprint` matches the trigger
-- **Notifications arrive late**: SigNoz sends grouped notifications every 5 minutes by default
+- **Resolved notifications arrive late**: SigNoz sends the resolved notification several minutes after the rule returns to normal, which is its default behavior

zh/on-call/integration/alert-integration/alert-sources/signoz.mdx

@@ -32,10 +32,10 @@ SigNoz 的 Webhook 通知渠道向你填写的 URL 发送 Prometheus Alertmanage
 ## 在 SigNoz 中配置
 ---
 
-1. 进入 SigNoz 的 **Settings** → **Account Settings** → **Notification Channels**，点击 **New Channel**
+1. 进入 SigNoz 的 **Settings** → **Notification Channels**（也可从 **Alerts** 页的 **Notification Channels** 标签进入），点击 **New Channel**
 2. 填写渠道名称，类型选择 **Webhook**，在 endpoint URL 中填入上一步复制的完整推送地址（包含 `?integration_key=...`）
 3. 保持 **Send resolved alerts** 开启（默认开启），否则 Flashduty 收不到恢复通知
-4. 保存，然后在 SigNoz 告警规则的通知渠道中选择该渠道
+4. 保存，然后在 SigNoz 告警规则的阈值行（**send to**）中选择该渠道
 
 ## 字段映射
 ---
@@ -50,18 +50,19 @@ SigNoz 的 Webhook 通知渠道向你填写的 URL 发送 Prometheus Alertmanage
 | 其余 `labels`、`annotations` | 每个键展开为一个标签，`annotations.description` 作为告警描述 |
 | `startsAt`、`endsAt` | 标签 `starts_at`、`ends_at` |
 
-SigNoz 按告警名称分组，每 5 分钟发送一次合并通知，一次通知可包含多条告警；Flashduty 逐条处理（单次最多 100 条）。
+SigNoz 按规则（`ruleId`）分组发送通知，一次通知可包含多条告警；Flashduty 逐条处理（单次最多 100 条）。规则触发通常在 1 分钟内送达。
 
 ## 恢复与去重
 ---
 
 - SigNoz 发送 `status` 为 `resolved` 的通知时，Flashduty 按相同的 `fingerprint` 关闭对应告警。
-- SigNoz 的严重程度是规则里的 `severity` 标签，取值由你定义。如果使用了 `error` 等未内置的取值，请在 Prometheus 集成中配置自定义严重程度映射，否则按 Warning 处理。
-- SigNoz 官方文档没有说明 Test 按钮发送的内容。首次配置后，用一条真实告警验证触发与恢复是否都出现在 Flashduty 中。
+- 新版 SigNoz 规则的 `severity` 标签取自阈值行的名称（例如阈值名 `critical`、`warning`），同时写入 `threshold.name` 标签。如果使用了 `error` 等未内置的取值，请在 Prometheus 集成中配置自定义严重程度映射，否则按 Warning 处理。
+- 恢复通知在规则回到正常后延迟数分钟到达（实测触发后约 30 秒、恢复约 6 至 10 分钟），属正常现象。
+- 渠道的 **Test** 按钮会发送一条 `status` 为 `firing` 的测试告警，标题为 `Test Alert (<渠道名称>)`，`severity` 为 `critical`。它会在 Flashduty 产生一条 Critical 告警，且不会有对应的恢复通知，请手动关闭该告警。
 
 ## 排查问题
 ---
 
 - **Flashduty 返回 `Invalid parameters`**：推送地址不完整，缺少 `integration_key`，或集成类型不是 Prometheus
 - **告警没有恢复**：确认 SigNoz 渠道的 **Send resolved alerts** 已开启，且恢复通知的 `fingerprint` 与触发时一致
-- **通知延迟**：SigNoz 每 5 分钟发送一次分组通知，这是 SigNoz 的默认行为
+- **恢复通知延迟**：SigNoz 在规则回到正常后延迟数分钟才发送恢复通知，这是 SigNoz 的默认行为