Refactor dashboard overview: remove active alerts and lagging nodes from summary, update tests and components accordingly

- Removed active alerts and lagging nodes from DashboardSummary and related types.
- Updated DashboardOverview component to reflect changes in data structure.
- Adjusted tests to align with the new dashboard overview structure, ensuring no active alerts are displayed.
- Simplified the dashboard metrics and risk signals, focusing on essential health and capacity metrics.
- Enhanced the WorldStage component to present updated traffic and capacity information.
- Removed unused alert-related functions and components to streamline the codebase.
This commit is contained in:
ryan
2026-03-14 16:00:41 +08:00
parent f33e9514dc
commit fe3c6312f9
8 changed files with 118 additions and 900 deletions
@@ -11,11 +11,7 @@ import { LoadingState } from '@/components/feedback/loading-state';
import { AppCard } from '@/components/ui/app-card';
import { StatusBadge } from '@/components/ui/status-badge';
import { getDashboardOverview } from '@/features/dashboard/api/overview';
import type {
DashboardAlert,
DashboardNodeHealth,
DashboardPeakNode,
} from '@/features/dashboard/types';
import type { DashboardNodeHealth } from '@/features/dashboard/types';
import { WorldStage } from '@/features/dashboard/components/world-stage';
import {
getNodeStatusLabel,
@@ -59,48 +55,6 @@ function formatTrendHour(value: string) {
return `${date.getHours().toString().padStart(2, '0')}:00`;
}
function formatPeakHour(value: string) {
if (!value) {
return '暂无';
}
const date = new Date(value);
if (Number.isNaN(date.getTime())) {
return '暂无';
}
return `${formatTrendHour(value)} - ${date
.getMinutes()
.toString()
.padStart(2, '0')}`;
}
function getAlertVariant(
alert: DashboardAlert,
): 'success' | 'warning' | 'danger' | 'info' {
if (alert.status === 'resolved') {
return 'success';
}
if (alert.severity === 'critical') {
return 'danger';
}
if (alert.severity === 'warning') {
return 'warning';
}
return 'info';
}
function formatPeakNode(node: DashboardPeakNode | null) {
if (!node) {
return {
title: '暂无',
hint: '当前没有可用节点数据',
};
}
return {
title: node.node_name,
hint: `请求 ${node.request_count} · 错误 ${node.error_count} · CPU ${formatPercent(node.cpu_usage_percent)}`,
};
}
function buildNodeDetailHref(id?: number | null) {
if (!id) {
return '/node';
@@ -173,82 +127,6 @@ function OverviewMetric({
);
}
function RiskSignal({
label,
value,
tone,
hint,
href,
}: {
label: string;
value: number;
tone: 'danger' | 'warning' | 'info' | 'success';
hint: string;
href?: string;
}) {
const toneClass =
tone === 'danger'
? 'border-rose-400/28 bg-rose-500/10 text-rose-100'
: tone === 'warning'
? 'border-amber-400/28 bg-amber-500/10 text-amber-100'
: tone === 'success'
? 'border-emerald-400/28 bg-emerald-500/10 text-emerald-100'
: 'border-sky-400/28 bg-sky-500/10 text-sky-100';
const content = (
<div className={`rounded-3xl border px-4 py-4 ${toneClass}`}>
<p className="text-xs tracking-[0.22em] opacity-75 uppercase">{label}</p>
<p className="mt-3 text-3xl font-semibold">{value}</p>
<p className="mt-2 text-sm opacity-80">{hint}</p>
</div>
);
if (!href) {
return content;
}
return (
<Link href={href} className="block transition hover:opacity-90">
{content}
</Link>
);
}
function PeakCard({
label,
value,
hint,
href,
}: {
label: string;
value: string;
hint: string;
href?: string;
}) {
const content = (
<div className="rounded-3xl border border-[var(--border-default)] bg-[var(--surface-elevated)] px-4 py-4">
<p className="text-xs tracking-[0.22em] text-[var(--foreground-muted)] uppercase">
{label}
</p>
<p className="mt-3 text-xl font-semibold text-[var(--foreground-primary)]">
{value}
</p>
<p className="mt-2 text-sm text-[var(--foreground-secondary)]">{hint}</p>
</div>
);
if (!href) {
return content;
}
return (
<Link
href={href}
className="block transition hover:[&_div]:border-[var(--border-strong)] hover:[&_div]:bg-[var(--surface-muted)]"
>
{content}
</Link>
);
}
function NodeHealthRow({ node }: { node: DashboardNodeHealth }) {
return (
<Link
@@ -354,148 +232,60 @@ export function DashboardOverview() {
);
}
const nodeIdMap = new Map(
overview.nodes.map((node) => [node.node_id, node.id] as const),
);
return (
<div className="space-y-6">
<WorldStage
generatedAt={overview.generated_at}
summary={overview.summary}
risk={overview.risk}
config={overview.config}
traffic={overview.traffic}
capacity={overview.capacity}
nodes={overview.nodes}
sourceCountries={overview.distributions.source_countries}
/>
<div className="grid gap-6 xl:grid-cols-[1.15fr_0.85fr]">
<AppCard
title="系统健康摘要"
description="先回答系统是否健康、容量是否紧张、是否存在追平偏差和异常流量。"
>
<div className="grid gap-4 md:grid-cols-2 xl:grid-cols-3">
<OverviewMetric
label="在线节点"
value={`${overview.summary.online_nodes}/${overview.summary.total_nodes}`}
hint={`${overview.summary.offline_nodes} 离线 · ${overview.summary.pending_nodes} 待接入`}
accent="sky"
/>
<OverviewMetric
label="活动异常"
value={overview.summary.active_alerts}
hint={`${overview.summary.unhealthy_nodes} 个 OpenResty 不健康`}
accent="rose"
/>
<OverviewMetric
label="配置落后"
value={overview.summary.lagging_nodes}
hint={overview.config.active_version || '当前无激活版本'}
accent="amber"
/>
<OverviewMetric
label="最近窗口请求"
value={overview.traffic.request_count.toLocaleString('zh-CN')}
hint={`QPS ${overview.traffic.estimated_qps.toFixed(1)} · UV ${overview.traffic.unique_visitors.toLocaleString('zh-CN')}`}
accent="emerald"
/>
<OverviewMetric
label="平均 CPU"
value={formatPercent(overview.capacity.average_cpu_usage_percent)}
hint={`${overview.capacity.high_cpu_nodes} 个高 CPU 节点`}
accent="amber"
/>
<OverviewMetric
label="平均内存"
value={formatPercent(overview.capacity.average_memory_usage_percent)}
hint={`${overview.capacity.high_memory_nodes} 个高内存节点`}
accent="sky"
/>
</div>
</AppCard>
<AppCard
title="峰值与风险摘要"
description="快速回答什么时候最忙、哪里最危险,以及首页能直接行动的入口。"
>
<div className="grid gap-4 md:grid-cols-2">
<PeakCard
label="请求峰值时段"
value={formatPeakHour(
overview.peaks.peak_request_hour.bucket_started_at,
)}
hint={`峰值请求 ${overview.peaks.peak_request_hour.request_count.toLocaleString('zh-CN')}`}
/>
<PeakCard
label="错误峰值时段"
value={formatPeakHour(
overview.peaks.peak_error_hour.bucket_started_at,
)}
hint={`峰值错误 ${overview.peaks.peak_error_hour.error_count.toLocaleString('zh-CN')}`}
/>
<PeakCard
label="最忙节点"
value={formatPeakNode(overview.peaks.busiest_node).title}
hint={formatPeakNode(overview.peaks.busiest_node).hint}
href={buildNodeDetailHref(
nodeIdMap.get(overview.peaks.busiest_node?.node_id ?? ''),
)}
/>
<PeakCard
label="优先排查节点"
value={formatPeakNode(overview.peaks.riskiest_node).title}
hint={formatPeakNode(overview.peaks.riskiest_node).hint}
href={buildNodeDetailHref(
nodeIdMap.get(overview.peaks.riskiest_node?.node_id ?? ''),
)}
/>
</div>
<div className="mt-4 grid gap-4 md:grid-cols-2">
<RiskSignal
label="Critical"
value={overview.risk.critical_alerts}
tone="danger"
hint="当前仍在触发中的严重异常"
/>
<RiskSignal
label="Warning"
value={overview.risk.warning_alerts}
tone="warning"
hint="需要尽快介入但尚未到故障级"
/>
<RiskSignal
label="配置落后"
value={overview.risk.lagging_nodes}
tone="info"
hint="当前版本未追平全局激活配置"
href="/node?risk=lagging"
/>
<RiskSignal
label="OpenResty 异常"
value={overview.risk.unhealthy_nodes}
tone="danger"
hint="运行态已出现不健康节点"
href="/node?risk=unhealthy"
/>
<RiskSignal
label="高 CPU / 内存"
value={
overview.risk.high_cpu_nodes + overview.risk.high_memory_nodes
}
tone="warning"
hint={`${overview.risk.high_cpu_nodes} 个高 CPU · ${overview.risk.high_memory_nodes} 个高内存`}
/>
<RiskSignal
label="离线节点"
value={overview.risk.offline_nodes}
tone="warning"
hint="节点长时间未心跳或已失联"
href="/node?risk=offline"
/>
</div>
</AppCard>
</div>
<AppCard
title="系统健康摘要"
description="聚焦在线率、运行质量、流量体量与资源负载,保留首屏最核心的运营信号。"
>
<div className="grid gap-4 md:grid-cols-2 xl:grid-cols-3">
<OverviewMetric
label="在线节点"
value={`${overview.summary.online_nodes}/${overview.summary.total_nodes}`}
hint={`${overview.summary.offline_nodes} 离线 · ${overview.summary.pending_nodes} 待接入`}
accent="sky"
/>
<OverviewMetric
label="OpenResty 异常"
value={overview.summary.unhealthy_nodes}
hint="当前运行态已出现不健康节点"
accent="rose"
/>
<OverviewMetric
label="最近窗口请求"
value={overview.traffic.request_count.toLocaleString('zh-CN')}
hint={`QPS ${overview.traffic.estimated_qps.toFixed(1)} · UV ${overview.traffic.unique_visitors.toLocaleString('zh-CN')}`}
accent="emerald"
/>
<OverviewMetric
label="平均 CPU"
value={formatPercent(overview.capacity.average_cpu_usage_percent)}
hint={`${overview.capacity.high_cpu_nodes} 个高 CPU 节点`}
accent="amber"
/>
<OverviewMetric
label="平均内存"
value={formatPercent(overview.capacity.average_memory_usage_percent)}
hint={`${overview.capacity.high_memory_nodes} 个高内存节点`}
accent="sky"
/>
<OverviewMetric
label="高存储节点"
value={overview.capacity.high_storage_nodes}
hint={`${overview.traffic.reported_nodes} 个节点参与当前窗口上报`}
accent="amber"
/>
</div>
</AppCard>
<div className="grid gap-6 xl:grid-cols-2">
<AppCard
@@ -559,7 +349,7 @@ export function DashboardOverview() {
</AppCard>
</div>
<div className="grid gap-6 xl:grid-cols-[1.15fr_0.85fr]">
<div className="grid gap-6 xl:grid-cols-[1.1fr_0.9fr]">
<AppCard
title="24 小时网络与磁盘趋势"
description="把 OpenResty 吞吐和磁盘读写放到同一屏,方便判断是流量拉升还是资源抖动。"
@@ -621,94 +411,9 @@ export function DashboardOverview() {
</div>
</AppCard>
<AppCard
title="活动异常"
description="优先展示当前仍在触发中的问题。"
>
{overview.active_alerts.length === 0 ? (
<EmptyState
title="暂无活动异常"
description="当前系统没有正在触发的节点健康事件。"
/>
) : (
<div className="space-y-3">
{overview.active_alerts.map((alert) => (
<Link
key={`${alert.node_id}-${alert.event_type}-${alert.last_triggered_at}`}
href={buildNodeDetailHref(nodeIdMap.get(alert.node_id))}
className="block rounded-2xl border border-[var(--border-default)] bg-[var(--surface-elevated)] px-4 py-4 transition hover:border-[var(--border-strong)] hover:bg-[var(--surface-muted)]"
>
<div className="flex flex-wrap items-center gap-2">
<StatusBadge
label={alert.event_type.replaceAll('_', ' ')}
variant={getAlertVariant(alert)}
/>
<p className="text-sm font-medium text-[var(--foreground-primary)]">
{alert.node_name}
</p>
</div>
<p className="mt-2 text-sm text-[var(--foreground-secondary)]">
{alert.message || '暂无详细消息'}
</p>
<p className="mt-2 text-xs text-[var(--foreground-muted)]">
{formatRelativeTime(alert.last_triggered_at)} ·{' '}
{formatDateTime(alert.last_triggered_at)}
</p>
</Link>
))}
</div>
)}
</AppCard>
</div>
<div className="grid gap-6 xl:grid-cols-3">
<AppCard
title="来源分布"
description="聚合最近 24 小时主要来源国家,优先识别流量重心变化。"
>
<RankChart
items={overview.distributions.source_countries.map((item) => ({
label: item.key,
value: item.value,
}))}
color="#38bdf8"
emptyMessage="暂无来源分布数据"
/>
</AppCard>
<AppCard
title="状态码分布"
description="快速判断成功响应是否仍是主流,以及错误码是否有抬升。"
>
<RankChart
items={overview.distributions.status_codes.map((item) => ({
label: `HTTP ${item.key}`,
value: item.value,
}))}
color="#f59e0b"
emptyMessage="暂无状态码分布"
/>
</AppCard>
<AppCard
title="Top Domain"
description="观察主要流量集中在哪些域名,方便判断业务热度与异常集中点。"
>
<RankChart
items={overview.distributions.top_domains.map((item) => ({
label: item.key,
value: item.value,
}))}
color="#34d399"
emptyMessage="暂无域名分布"
/>
</AppCard>
</div>
<div className="grid gap-6 xl:grid-cols-[1.2fr_0.8fr]">
<AppCard
title="Top 节点榜单"
description="把流量峰值、错误热点和容量压力节点并排展示,方便值守时快速判断先处理谁。"
description="把流量、错误与容量压力节点并排展示,保留快速定位热点与瓶颈的入口。"
>
<div className="grid gap-6 xl:grid-cols-3">
<div>
@@ -760,61 +465,49 @@ export function DashboardOverview() {
</div>
</div>
</AppCard>
</div>
<div className="grid gap-6 xl:grid-cols-3">
<AppCard
title="来源分布"
description="聚合最近 24 小时主要来源国家,优先识别流量重心变化。"
>
<RankChart
items={overview.distributions.source_countries.map((item) => ({
label: item.key,
value: item.value,
}))}
color="#38bdf8"
emptyMessage="暂无来源分布数据"
/>
</AppCard>
<AppCard
title="处置建议"
description="把首页信号翻译成可行动的下一步,降低值守判断成本。"
title="状态码分布"
description="快速判断成功响应是否仍是主流,以及错误码是否有抬升。"
>
<div className="space-y-4">
<div className="rounded-3xl border border-rose-400/18 bg-rose-500/8 px-4 py-4">
<p className="text-xs tracking-[0.22em] text-rose-200 uppercase">
第一优先级
</p>
<p className="mt-3 text-lg font-semibold text-[var(--foreground-primary)]">
先处理活动异常与 OpenResty 不健康节点
</p>
<p className="mt-2 text-sm text-[var(--foreground-secondary)]">
当前有 {overview.summary.active_alerts} 个活动异常,{` `}
{overview.summary.unhealthy_nodes} 个 OpenResty 不健康节点。
</p>
</div>
<RankChart
items={overview.distributions.status_codes.map((item) => ({
label: `HTTP ${item.key}`,
value: item.value,
}))}
color="#f59e0b"
emptyMessage="暂无状态码分布"
/>
</AppCard>
<div className="rounded-3xl border border-amber-400/18 bg-amber-500/8 px-4 py-4">
<p className="text-xs tracking-[0.22em] text-amber-200 uppercase">
第二优先级
</p>
<p className="mt-3 text-lg font-semibold text-[var(--foreground-primary)]">
检查配置落后与容量逼近节点
</p>
<p className="mt-2 text-sm text-[var(--foreground-secondary)]">
当前有 {overview.summary.lagging_nodes} 个配置落后节点,{` `}
{overview.capacity.high_cpu_nodes +
overview.capacity.high_memory_nodes +
overview.capacity.high_storage_nodes}{' '}
个节点出现明显容量压力。
</p>
</div>
<div className="rounded-3xl border border-sky-400/18 bg-sky-500/8 px-4 py-4">
<p className="text-xs tracking-[0.22em] text-sky-200 uppercase">
观察项
</p>
<p className="mt-3 text-lg font-semibold text-[var(--foreground-primary)]">
关注峰值时段与来源分布变化
</p>
<p className="mt-2 text-sm text-[var(--foreground-secondary)]">
请求峰值出现在 {formatPeakHour(
overview.peaks.peak_request_hour.bucket_started_at,
)}
,当前主要来源集中在
{overview.distributions.source_countries
.slice(0, 2)
.map((item) => item.key)
.join(' / ') || ' 暂无来源数据'}
。
</p>
</div>
</div>
<AppCard
title="Top Domain"
description="观察主要流量集中在哪些域名,方便判断业务热度与异常集中点。"
>
<RankChart
items={overview.distributions.top_domains.map((item) => ({
label: item.key,
value: item.value,
}))}
color="#34d399"
emptyMessage="暂无域名分布"
/>
</AppCard>
</div>