feat: expand resource diagnostics

CHANGELOG.md

@@ -8,6 +8,8 @@ Follows [Keep a Changelog](https://keepachangelog.com/en/1.0.0/) · versioned wi
 ## [Unreleased]
 
 ### Changed
+- **Expanded resource diagnostics.** The Performance panel now reports bounded browser cache counters, container
+  cgroup current/max memory, artifact-cache bytes, and sanitized largest recent operation durations.
 - **Current-release constrained validation.** Built `1.24.21` and served it under a hard 4 GiB cgroup limit;
   startup remained healthy at 52.54 MiB, 0.10% CPU, 6 PIDs, and no OOM kill.
 - **Release cache-busting.** Updated the HTML app-version, stylesheet URL, and loader URL to match `VERSION`

TODO.md

@@ -229,8 +229,9 @@ architecture change: the larger Audiobook/Rehearser/style split.
   ~50.6 MiB idle memory, 0.09% CPU, and no OOM kill.
 - [x] Monitor browser-tab memory, JavaScript heap where supported, DOM-node count, audio elements, persisted jobs,
   server RSS, configured-data disk, and temporary disk usage through the read-only Performance diagnostics panel.
-- [ ] Add a memory diagnostics page showing cache count/size, Docker/container memory, and largest recent operations;
-  the current panel covers browser/server/temp basics without pretending to measure unavailable host-level metrics.
+- [x] Add a memory diagnostics page showing cache count/size, Docker/container memory, and largest recent operations.
+  The authenticated panel reports browser cache counters, cgroup memory current/max, server RSS, configured/temp
+  storage, persisted jobs, and sanitized operation durations; host-wide Docker metrics remain unavailable by design.
 
 ## P1 — Speaker attribution: measured state and next steps (paused 2026-09-02)
 

VERSION

@@ -1 +1 @@
-1.24.22
+1.24.23

routes/jobs.py

@@ -5,6 +5,7 @@ import time
 import resource
 import shutil
 from pathlib import Path
+from typing import Any
 
 from fastapi import APIRouter, HTTPException
 from fastapi.responses import FileResponse
@@ -28,6 +29,36 @@ def _directory_bytes(root: Path) -> int:
     return total
 
 
+def _cgroup_bytes(name: str) -> int | None:
+    """Read the container's cgroup v2 memory metric when available."""
+    try:
+        value = (Path('/sys/fs/cgroup') / name).read_text(encoding='utf-8').strip()
+        if value == 'max': return None
+        return max(0, int(value))
+    except (OSError, ValueError):
+        return None
+
+
+def _largest_recent_jobs(limit: int = 5) -> list[dict[str, Any]]:
+    rows = []
+    now = time.time()
+    for job in background_job_list(limit=100):
+        if not job: continue
+        try:
+            started = float(job.get('created') or now)
+            finished = float(job.get('updated') or now)
+            elapsed = max(0.0, finished - started)
+        except (TypeError, ValueError):
+            continue
+        rows.append({
+            'kind': str(job.get('kind') or 'job'),
+            'job_id': str(job.get('job_id') or '')[:12],
+            'elapsed': round(elapsed, 1),
+            'status': 'failed' if job.get('error') else ('completed' if job.get('done') else 'running'),
+        })
+    return sorted(rows, key=lambda row: row['elapsed'], reverse=True)[:max(1, min(10, limit))]
+
+
 def _public_job(job: dict) -> dict:
     payload = dict(job)
     payload.pop("fingerprint", None)
@@ -74,11 +105,18 @@ async def jobs_list(kind: str = "", limit: int = 50):
 @router.get("/api/diagnostics")
 async def diagnostics():
     usage = shutil.disk_usage(str(CONFIG_DIR))
+    artifacts = CONFIG_DIR / 'background-artifacts'
     return {
         "server_rss_mb": round(float(resource.getrusage(resource.RUSAGE_SELF).ru_maxrss) / 1024, 1),
         "config_disk": {"used_bytes": _directory_bytes(CONFIG_DIR), "free_bytes": usage.free},
         "temp_disk": {"used_bytes": _directory_bytes(TEMP_DIR)},
         "background_jobs": len(background_job_list(limit=100)),
+        "container_memory": {
+            "current_bytes": _cgroup_bytes('memory.current'),
+            "max_bytes": _cgroup_bytes('memory.max'),
+        },
+        "cache_storage": {"background_artifacts_bytes": _directory_bytes(artifacts)},
+        "largest_operations": _largest_recent_jobs(),
     }
 
 

static/js/jobs.js

@@ -159,11 +159,17 @@
         domNodes: document.getElementsByTagName('*').length,
         audioElements: document.querySelectorAll('audio').length,
         jsHeapBytes: performance.memory?.usedJSHeapSize || null,
+        imageCacheEntries: window.TTVC?.imageCache?.size?.() ?? null,
+        rehearserAudioEntries: window.rehState?.synthCache?.size ?? null,
+        voiceRecords: Array.isArray(window._voices) ? window._voices.length : null,
       })));
       diagnostics.replaceChildren();
       [["Server RSS", `${server.server_rss_mb} MiB`], ["Config data", formatBytes(server.config_disk?.used_bytes)],
         ["Config free", formatBytes(server.config_disk?.free_bytes)], ["Temp data", formatBytes(server.temp_disk?.used_bytes)],
-        ["Persisted jobs", server.background_jobs], ["DOM nodes", browser.domNodes],
+        ["Container memory", `${formatBytes(server.container_memory?.current_bytes)} / ${formatBytes(server.container_memory?.max_bytes)}`],
+        ["Artifact cache", formatBytes(server.cache_storage?.background_artifacts_bytes)], ["Persisted jobs", server.background_jobs],
+        ["Image cache entries", browser.imageCacheEntries], ["Rehearser audio entries", browser.rehearserAudioEntries],
+        ["Voice records", browser.voiceRecords], ["DOM nodes", browser.domNodes],
         ["Audio elements", browser.audioElements], ["JS heap", formatBytes(browser.jsHeapBytes)]].forEach(([label, value]) => {
         const item = document.createElement('div');
         item.className = 'resource-diagnostic-item';
@@ -171,6 +177,17 @@
         const valueNode = document.createElement('strong'); valueNode.textContent = value == null ? '—' : String(value);
         item.append(name, valueNode); diagnostics.appendChild(item);
       });
+      const operations = Array.isArray(server.largest_operations) ? server.largest_operations : [];
+      if (operations.length) {
+        const heading = document.createElement('div'); heading.className = 'resource-diagnostic-subhead'; heading.textContent = 'Largest recent operations';
+        diagnostics.appendChild(heading);
+        operations.forEach(operation => {
+          const item = document.createElement('div'); item.className = 'resource-diagnostic-item';
+          const name = document.createElement('span'); name.textContent = `${operation.kind} · ${operation.status}`;
+          const valueNode = document.createElement('strong'); valueNode.textContent = `${Number(operation.elapsed || 0).toFixed(1)}s`;
+          item.append(name, valueNode); diagnostics.appendChild(item);
+        });
+      }
     } catch (error) {
       diagnostics.replaceChildren();
       const item = node('Diagnostics unavailable: ' + error.message); item.className = 'perf-history-empty'; diagnostics.appendChild(item);

static/style.css

@@ -3746,6 +3746,7 @@ code { background: var(--panel); border-radius: 4px; padding: 1px 5px; font-fami
 .resource-diagnostic-item { display: flex; flex-direction: column; gap: 3px; padding: 9px 10px; border: 1px solid var(--border); border-radius: 6px; background: var(--panel); }
 .resource-diagnostic-item span { color: var(--subtext); font-size: 11px; }
 .resource-diagnostic-item strong { font-size: 14px; }
+.resource-diagnostic-subhead { grid-column: 1 / -1; margin-top: 8px; color: var(--subtext); font-size: 12px; font-weight: 700; }
 .background-job-row.is-error .background-job-detail { color: var(--red); }
 .background-job-row.is-done .background-job-meta { color: var(--green, #22c55e); }
 .background-job-actions { display: flex; align-items: center; gap: 9px; flex-shrink: 0; }

tests/smoke.spec.js

@@ -109,3 +109,22 @@ test('background job history exposes retry only when the server allows it', asyn
   expect(clicked).toBeTruthy();
   await expect.poll(() => retryRequests).toBe(1);
 });
+
+test('resource diagnostics expose cache, container, and operation metrics', async ({ page }) => {
+  await page.route('**/api/diagnostics', route => route.fulfill({ json: {
+    server_rss_mb: 52.5,
+    config_disk: { used_bytes: 1024, free_bytes: 2048 },
+    temp_disk: { used_bytes: 512 },
+    background_jobs: 2,
+    container_memory: { current_bytes: 4096, max_bytes: 8192 },
+    cache_storage: { background_artifacts_bytes: 256 },
+    largest_operations: [{ kind: 'voice_design', status: 'completed', elapsed: 4.2 }],
+  } }));
+  await page.goto('/', { waitUntil: 'domcontentloaded' });
+  await navTo(page, 's-performance');
+  const diagnostics = page.locator('#resource-diagnostics');
+  await expect(diagnostics).toContainText('Image cache entries');
+  await expect(diagnostics).toContainText('Container memory');
+  await expect(diagnostics).toContainText('Largest recent operations');
+  await expect(diagnostics).toContainText('voice_design · completed');
+});