Refactor Activity Page (#710)

- inference handles to store an activity record for all inference endpoints
- add path, status code, and content type to Activities page
- toggle on/off columns no Activities page 
- add configurable capture level for inference endpoints so large binary blobs are not stored in memory
- store captures in compressed binary format
This commit is contained in:
Benson Wong
2026-04-28 20:33:03 -07:00
committed by GitHub
parent a846c4f18c
commit fd3c28ffc5
16 changed files with 1397 additions and 651 deletions
+3 -3
View File
@@ -2788,9 +2788,9 @@
}
},
"node_modules/postcss": {
"version": "8.5.8",
"resolved": "https://registry.npmjs.org/postcss/-/postcss-8.5.8.tgz",
"integrity": "sha512-OW/rX8O/jXnm82Ey1k44pObPtdblfiuWnrd8X7GJ7emImCOstunGbXUpp7HdBrFQX6rJzn3sPT397Wp5aCwCHg==",
"version": "8.5.12",
"resolved": "https://registry.npmjs.org/postcss/-/postcss-8.5.12.tgz",
"integrity": "sha512-W62t/Se6rA0Az3DfCL0AqJwXuKwBeYg6nOaIgzP+xZ7N5BFCI7DYi1qs6ygUYT6rvfi6t9k65UMLJC+PHZpDAA==",
"dev": true,
"funding": [
{
@@ -9,13 +9,13 @@
let stats = $derived.by(() => {
const totalRequests = $metrics.length;
const totalInputTokens = $metrics.reduce((sum, m) => sum + m.input_tokens, 0);
const totalOutputTokens = $metrics.reduce((sum, m) => sum + m.output_tokens, 0);
const totalCacheTokens = $metrics.reduce((sum, m) => sum + m.cache_tokens, 0);
const totalInputTokens = $metrics.reduce((sum, m) => sum + m.tokens.input_tokens, 0);
const totalOutputTokens = $metrics.reduce((sum, m) => sum + m.tokens.output_tokens, 0);
const totalCacheTokens = $metrics.reduce((sum, m) => sum + m.tokens.cache_tokens, 0);
const promptPerSecond = $metrics.filter((m) => m.prompt_per_second > 0).map((m) => m.prompt_per_second);
const promptPerSecond = $metrics.filter((m) => m.tokens.prompt_per_second > 0).map((m) => m.tokens.prompt_per_second);
const tokensPerSecond = $metrics.filter((m) => m.tokens_per_second > 0).map((m) => m.tokens_per_second);
const tokensPerSecond = $metrics.filter((m) => m.tokens.tokens_per_second > 0).map((m) => m.tokens.tokens_per_second);
const promptHistogramData =
promptPerSecond.length > 0 ? calculateHistogramData(promptPerSecond) : null;
+11 -4
View File
@@ -12,15 +12,22 @@ export interface Model {
aliases?: string[];
}
export interface Metrics {
id: number;
timestamp: string;
model: string;
export interface TokenMetrics {
cache_tokens: number;
input_tokens: number;
output_tokens: number;
prompt_per_second: number;
tokens_per_second: number;
}
export interface ActivityLogEntry {
id: number;
timestamp: string;
model: string;
req_path: string;
resp_content_type: string;
resp_status_code: number;
tokens: TokenMetrics;
duration_ms: number;
has_capture: boolean;
}
+209 -38
View File
@@ -3,8 +3,87 @@
import ActivityStats from "../components/ActivityStats.svelte";
import Tooltip from "../components/Tooltip.svelte";
import CaptureDialog from "../components/CaptureDialog.svelte";
import { persistentStore } from "../stores/persistent";
import { onMount } from "svelte";
import type { ReqRespCapture } from "../lib/types";
type ColumnKey =
| "id"
| "time"
| "model"
| "req_path"
| "resp_status_code"
| "resp_content_type"
| "cached"
| "prompt"
| "generated"
| "prompt_speed"
| "gen_speed"
| "duration"
| "capture";
interface ColumnDef {
key: ColumnKey;
label: string;
defaultVisible: boolean;
}
const columns: ColumnDef[] = [
{ key: "id", label: "ID", defaultVisible: true },
{ key: "time", label: "Time", defaultVisible: true },
{ key: "model", label: "Model", defaultVisible: true },
{ key: "req_path", label: "Path", defaultVisible: false },
{ key: "resp_status_code", label: "Status", defaultVisible: false },
{ key: "resp_content_type", label: "Content-Type", defaultVisible: false },
{ key: "cached", label: "Cached", defaultVisible: true },
{ key: "prompt", label: "Prompt", defaultVisible: true },
{ key: "generated", label: "Generated", defaultVisible: true },
{ key: "prompt_speed", label: "Prompt Speed", defaultVisible: true },
{ key: "gen_speed", label: "Gen Speed", defaultVisible: true },
{ key: "duration", label: "Duration", defaultVisible: true },
{ key: "capture", label: "Capture", defaultVisible: true },
];
const defaultVisibleKeys = columns.filter((c) => c.defaultVisible).map((c) => c.key);
const visibleColumns = persistentStore<ColumnKey[]>(
"activity-columns",
defaultVisibleKeys
);
let columnsMenuOpen = $state(false);
let dropdownContainer: HTMLDivElement | null = null;
onMount(() => {
function handleKeydown(e: KeyboardEvent) {
if (e.key === "Escape" && columnsMenuOpen) {
columnsMenuOpen = false;
}
}
function handleClick(e: MouseEvent) {
if (columnsMenuOpen && dropdownContainer && !dropdownContainer.contains(e.target as Node)) {
columnsMenuOpen = false;
}
}
document.addEventListener("keydown", handleKeydown);
document.addEventListener("click", handleClick);
return () => {
document.removeEventListener("keydown", handleKeydown);
document.removeEventListener("click", handleClick);
};
});
function toggleColumn(key: ColumnKey) {
const current = $visibleColumns;
if (current.includes(key)) {
if (current.length > 1) {
visibleColumns.set(current.filter((k) => k !== key));
}
} else {
visibleColumns.set([...current, key]);
}
}
function formatSpeed(speed: number): string {
return speed < 0 ? "unknown" : speed.toFixed(2) + " t/s";
}
@@ -67,58 +146,150 @@
<ActivityStats />
</div>
<div class="card overflow-auto">
<div class="card overflow-auto relative min-h-[30rem]">
<div class="flex justify-end px-4" bind:this={dropdownContainer}>
<div class="relative">
<button
class="w-8 h-8 flex items-center justify-center rounded hover:bg-secondary-hover transition-colors"
onclick={() => (columnsMenuOpen = !columnsMenuOpen)}
title="Select columns"
>
<svg class="w-4 h-4" fill="none" stroke="currentColor" viewBox="0 0 24 24">
<path stroke-linecap="round" stroke-linejoin="round" stroke-width="2" d="M12 6V4m0 2a2 2 0 100 4m0-4a2 2 0 110 4m-6 8a2 2 0 100-4m0 4a2 2 0 110-4m0 4v2m0-6V4m6 6v10m6-2a2 2 0 100-4m0 4a2 2 0 110-4m0 4v2m0-6V4"></path>
</svg>
</button>
{#if columnsMenuOpen}
<div class="absolute right-0 top-full mt-1 bg-surface border border-gray-200 dark:border-white/10 rounded shadow-lg z-10 py-1 min-w-[16rem]">
<div class="px-3 py-2 text-xs font-medium uppercase tracking-wider text-gray-500 dark:text-gray-400 border-b border-gray-200 dark:border-white/10">
Columns
</div>
{#each columns as col (col.key)}
<label
class="flex items-center gap-2 px-3 py-1.5 text-sm cursor-pointer hover:bg-secondary-hover transition-colors"
>
<input
type="checkbox"
checked={$visibleColumns.includes(col.key)}
onchange={() => toggleColumn(col.key)}
class="rounded"
/>
{col.label}
</label>
{/each}
</div>
{/if}
</div>
</div>
<table class="min-w-full divide-y">
<thead class="border-gray-200 dark:border-white/10">
<tr class="text-left text-xs uppercase tracking-wider">
<th class="px-6 py-3">ID</th>
<th class="px-6 py-3">Time</th>
<th class="px-6 py-3">Model</th>
<th class="px-6 py-3">
Cached <Tooltip content="prompt tokens from cache" />
</th>
<th class="px-6 py-3">
Prompt <Tooltip content="new prompt tokens processed" />
</th>
<th class="px-6 py-3">Generated</th>
<th class="px-6 py-3">Prompt Processing</th>
<th class="px-6 py-3">Generation Speed</th>
<th class="px-6 py-3">Duration</th>
<th class="px-6 py-3">Capture</th>
{#if $visibleColumns.includes("id")}
<th class="px-6 py-3">ID</th>
{/if}
{#if $visibleColumns.includes("time")}
<th class="px-6 py-3">Time</th>
{/if}
{#if $visibleColumns.includes("model")}
<th class="px-6 py-3">Model</th>
{/if}
{#if $visibleColumns.includes("req_path")}
<th class="px-6 py-3">Path</th>
{/if}
{#if $visibleColumns.includes("resp_status_code")}
<th class="px-6 py-3">Status</th>
{/if}
{#if $visibleColumns.includes("resp_content_type")}
<th class="px-6 py-3">Content-Type</th>
{/if}
{#if $visibleColumns.includes("cached")}
<th class="px-6 py-3">
Cached <Tooltip content="prompt tokens from cache" />
</th>
{/if}
{#if $visibleColumns.includes("prompt")}
<th class="px-6 py-3">
Prompt <Tooltip content="new prompt tokens processed" />
</th>
{/if}
{#if $visibleColumns.includes("generated")}
<th class="px-6 py-3">Generated</th>
{/if}
{#if $visibleColumns.includes("prompt_speed")}
<th class="px-6 py-3">Prompt Speed</th>
{/if}
{#if $visibleColumns.includes("gen_speed")}
<th class="px-6 py-3">Gen Speed</th>
{/if}
{#if $visibleColumns.includes("duration")}
<th class="px-6 py-3">Duration</th>
{/if}
{#if $visibleColumns.includes("capture")}
<th class="px-6 py-3">Capture</th>
{/if}
</tr>
</thead>
<tbody class="divide-y">
{#if sortedMetrics.length === 0}
<tr>
<td colspan="10" class="px-6 py-8 text-center text-sm text-gray-500 dark:text-gray-400">
<td colspan={$visibleColumns.length} class="px-6 py-8 text-center text-sm text-gray-500 dark:text-gray-400">
No activity recorded
</td>
</tr>
{:else}
{#each sortedMetrics as metric (metric.id)}
<tr class="whitespace-nowrap text-sm border-gray-200 dark:border-white/10">
<td class="px-4 py-4">{metric.id + 1}</td>
<td class="px-6 py-4">{formatRelativeTime(metric.timestamp)}</td>
<td class="px-6 py-4">{metric.model}</td>
<td class="px-6 py-4">{metric.cache_tokens > 0 ? metric.cache_tokens.toLocaleString() : "-"}</td>
<td class="px-6 py-4">{metric.input_tokens.toLocaleString()}</td>
<td class="px-6 py-4">{metric.output_tokens.toLocaleString()}</td>
<td class="px-6 py-4">{formatSpeed(metric.prompt_per_second)}</td>
<td class="px-6 py-4">{formatSpeed(metric.tokens_per_second)}</td>
<td class="px-6 py-4">{formatDuration(metric.duration_ms)}</td>
<td class="px-6 py-4">
{#if metric.has_capture}
<button
onclick={() => viewCapture(metric.id)}
disabled={loadingCaptureId === metric.id}
class="btn btn--sm"
>
{loadingCaptureId === metric.id ? "..." : "View"}
</button>
{:else}
<span class="text-txtsecondary">-</span>
{/if}
</td>
{#if $visibleColumns.includes("id")}
<td class="px-4 py-4">{metric.id + 1}</td>
{/if}
{#if $visibleColumns.includes("time")}
<td class="px-6 py-4">{formatRelativeTime(metric.timestamp)}</td>
{/if}
{#if $visibleColumns.includes("model")}
<td class="px-6 py-4">{metric.model}</td>
{/if}
{#if $visibleColumns.includes("req_path")}
<td class="px-6 py-4">{metric.req_path || "-"}</td>
{/if}
{#if $visibleColumns.includes("resp_status_code")}
<td class="px-6 py-4">{metric.resp_status_code || "-"}</td>
{/if}
{#if $visibleColumns.includes("resp_content_type")}
<td class="px-6 py-4">{metric.resp_content_type || "-"}</td>
{/if}
{#if $visibleColumns.includes("cached")}
<td class="px-6 py-4">{metric.tokens.cache_tokens > 0 ? metric.tokens.cache_tokens.toLocaleString() : "-"}</td>
{/if}
{#if $visibleColumns.includes("prompt")}
<td class="px-6 py-4">{metric.tokens.input_tokens.toLocaleString()}</td>
{/if}
{#if $visibleColumns.includes("generated")}
<td class="px-6 py-4">{metric.tokens.output_tokens.toLocaleString()}</td>
{/if}
{#if $visibleColumns.includes("prompt_speed")}
<td class="px-6 py-4">{formatSpeed(metric.tokens.prompt_per_second)}</td>
{/if}
{#if $visibleColumns.includes("gen_speed")}
<td class="px-6 py-4">{formatSpeed(metric.tokens.tokens_per_second)}</td>
{/if}
{#if $visibleColumns.includes("duration")}
<td class="px-6 py-4">{formatDuration(metric.duration_ms)}</td>
{/if}
{#if $visibleColumns.includes("capture")}
<td class="px-6 py-4">
{#if metric.has_capture}
<button
onclick={() => viewCapture(metric.id)}
disabled={loadingCaptureId === metric.id}
class="btn btn--sm"
>
{loadingCaptureId === metric.id ? "..." : "View"}
</button>
{:else}
<span class="text-txtsecondary">-</span>
{/if}
</td>
{/if}
</tr>
{/each}
{/if}
+2 -2
View File
@@ -10,7 +10,7 @@
const viewModeStore = persistentStore<ViewMode>("logviewer-view-mode", "panels");
let direction = $derived<"horizontal" | "vertical">(
$screenWidth === "xs" || $screenWidth === "sm" ? "vertical" : "horizontal"
$screenWidth === "xs" || $screenWidth === "sm" ? "vertical" : "horizontal",
);
</script>
@@ -30,7 +30,7 @@
class:bg-primary={$viewModeStore === "proxy"}
class:text-btn-primary-text={$viewModeStore === "proxy"}
>
Panel
Proxy
</button>
<button
onclick={() => viewModeStore.set("upstream")}
+12 -4
View File
@@ -1,5 +1,13 @@
import { writable } from "svelte/store";
import type { Model, Metrics, VersionInfo, LogData, APIEventEnvelope, ReqRespCapture, InFlightStats } from "../lib/types";
import type {
Model,
ActivityLogEntry,
VersionInfo,
LogData,
APIEventEnvelope,
ReqRespCapture,
InFlightStats,
} from "../lib/types";
import { connectionState } from "./theme";
const LOG_LENGTH_LIMIT = 1024 * 100; /* 100KB of log data */
@@ -8,7 +16,7 @@ const LOG_LENGTH_LIMIT = 1024 * 100; /* 100KB of log data */
export const models = writable<Model[]>([]);
export const proxyLogs = writable<string>("");
export const upstreamLogs = writable<string>("");
export const metrics = writable<Metrics[]>([]);
export const metrics = writable<ActivityLogEntry[]>([]);
export const inFlightRequests = writable<number>(0);
export const versionInfo = writable<VersionInfo>({
build_date: "unknown",
@@ -62,7 +70,7 @@ export function enableAPIEvents(enabled: boolean): void {
const newModels = JSON.parse(message.data) as Model[];
// Sort models by name and id
newModels.sort((a, b) => {
return (a.name + a.id).localeCompare(b.name + b.id, undefined, { numeric : true} );
return (a.name + a.id).localeCompare(b.name + b.id, undefined, { numeric: true });
});
models.set(newModels);
break;
@@ -82,7 +90,7 @@ export function enableAPIEvents(enabled: boolean): void {
}
case "metrics": {
const newMetrics = JSON.parse(message.data) as Metrics[];
const newMetrics = JSON.parse(message.data) as ActivityLogEntry[];
metrics.update((prevMetrics) => [...newMetrics, ...prevMetrics]);
break;
}