Preserve guide data across EPG refresh failures

[cbulock]

Summary

• retain last-known-good XMLTV per source during transient fetch or file failures
• report explicit ok, degraded, and ailed EPG refresh outcomes through HTTP, scheduler, and MCP
• serialize refresh requests and follow an in-flight pass with the latest requested configuration
• remove the duplicate module interval; scheduler owns periodic refreshes

Validation

• �slint . passes
• 193 non-integration tests pass
• est/integration/guide.test.js passes (7 tests), including stale fallback, recovery, and all-source failure
• git diff --check passes

[Copilot]

🟡 Changes recommended

Some newly introduced outcome messaging and error-wrapping behavior is misleading or drops important error metadata, which can impair operational diagnosis and client handling.

Once you've addressed the issues Copilot identified, you can request another Copilot review.

Pull request overview

This PR improves EPG refresh resiliency by retaining last-known-good XMLTV per source when a fetch/read fails, and by surfacing explicit refresh outcomes (ok, degraded, failed) through the scheduler, HTTP reload endpoint, and MCP tool.

Changes:

• Serialize concurrent EPG refresh requests and return a structured refresh outcome from refreshEPG().
• Retain and reuse last-good per-source XMLTV data to keep guide output available during transient source failures, and remove the module-owned periodic interval in favor of the scheduler job.
• Expand integration coverage to validate stale fallback, recovery, and all-source failure behavior.

File summaries

File	Description
test/integration/guide.test.js	Adds integration coverage for stale fallback, recovery, and full failure outcomes.
server/epg.js	Implements serialized refresh orchestration, per-source last-good retention, and refresh outcome reporting.
server/scheduler.js	Treats failed refresh outcomes as job failures while allowing degraded runs to succeed.
server/mcp.js	Returns explicit MCP error on failed outcome and includes outcome details in reload_epg responses.
server/config.js	Updates /api/reload/epg to return outcome status and per-source results to callers.

Review details

Suppressed comments (1)

server/epg.js:430

• Wrapping the Axios error in a new Error drops Axios-specific fields (e.g., err.code / err.response.status), so the downstream catch block’s actionable logging branches won’t run when there is no last-known-good fallback.

          } catch (httpErr) {
            sourceFetchError = new Error(`Failed to fetch EPG: ${httpErr.message}`);
            const lastGood = lastGoodSourceXml.get(sourceName);
            if (!lastGood || lastGood.url !== sourceUrl) {
              throw sourceFetchError;
            }

• Files reviewed: 5/5 changed files
• Comments generated: 4
• Review effort level: Lite

---

💡 Add a code-review agent skill or configure MCP servers for context-aware, tailored reviews. Learn more in the docs.