Diagnostics resource timeline stays empty until manually refreshed #30520

What happened

In the Linux desktop app, Diagnostics > Resource timeline showed an empty chart and 'No retained process samples in this window.' Leaving Diagnostics open for a couple of minutes did not update it. Clicking the timeline refresh arrow immediately populated the chart and process table. Expected: the timeline updates as samples arrive, or clearly explains that manual refresh is required.

Diagnosis

Installed bundle inspection: the page subscribes to live resourceTelemetry, but resourceTelemetryHistory uses a separate query with staleTimeMs: 5000. The timeline refresh button calls that query's refresh. No automatic history-refresh timer was found in the diagnostics component. Native sampling uses interval 0 with no live subscribers. Likely interaction: an initially empty history result remains displayed after live collection starts. This is a source-based explanation, not a proven first-load race. The installed monitor exists, is executable, and emitted a valid protocol-v3 hello. A pinned source clone was unavailable because sandbox DNS could not resolve GitHub; evidence comes from the installed release bundle.

Steps to reproduce

1. Open the desktop app's Diagnostics for the local environment.
2. Observe an empty Resource timeline (15m selected).
3. Leave Diagnostics open for a couple of minutes: chart and table remain empty.
4. Click the Resource timeline refresh arrow beside 1h.
5. Chart and process rows immediately populate.
   Initial conditions causing the empty first result have not been isolated.

Version

0.1.7-nightly.20261103.4119 (affected desktop and local server). Triage CLI: same version.

Environment

Linux x64, kernel 7.2.15-1-omarchy; T3 Code Nightly desktop with local server. Node reported by triage: v24.22.1.

Evidence

User provided before/after screenshots and confirmed waiting several minutes, then successful manual refresh. After refresh, server and monitor rows each showed 336 samples. Trace spans ResourceTelemetry.readHistory and ws.rpc.server.getResourceTelemetryHistory completed successfully. Installed files: _chat.settings.diagnostics-BeChEshR.js, runtime-COJDfKd5.js, ManagedEndpointRuntime-Ck1eaot7.mjs.

Related issues

No matching duplicate found in web search. Related context: #11222 and #19399 changed idle sampling; neither establishes this stale-history UI issue. Newer nightly 0.1.7-nightly.20261103.4138 release notes list no matching fix; not retested there.

Fix applied or workaround

Click the Resource timeline refresh arrow after leaving Diagnostics open long enough to collect samples.

Filed by

Prepared by OpenAI GPT-6 through the Codex agent during T3 Code triage.