Android Telemetry Redaction FAQ (AND7)

This backlog captures the recurring questions that surfaced during the AND7 telemetry enablement sessions, governance reviews, and chaos rehearsals. It is the companion to the policy (telemetry_redaction.md), quick-reference card (telemetry_redaction_quick_reference.md), pre-read (telemetry_redaction_pre_read.md), and meeting minutes (telemetry_redaction_minutes_20260212.md). Update this FAQ after each workshop or incident so on-call staff have a single canonical thread for escalation guidance.

When should I request a redaction override?

Overrides are the last resort when a pipeline fault leaves data stuck in the redaction queue and threatens SLAs. File an override only after:

1. Running the chaos checklist (telemetry_chaos_checklist.md) steps that apply to the failing signal.
2. Capturing CLI evidence from scripts/telemetry/inject_redaction_failure.sh and scripts/telemetry/check_redaction_status.py.
3. Confirming with the Android Observability TL that the override is scoped to a single incident window.

Once approved, record the override in the audit log (telemetry_override_log.md) and include the token hash plus expiry timestamp. The quick-reference card lists the exact CLI invocation for both the approval and cleanup commands.

What evidence is required before closing an incident?

Collect the following artefacts and attach them to the incident ticket before marking an override resolved:

• Latest status bundle JSON (artifacts/android/telemetry/status-*.json).
• Grafana screenshots highlighting android.telemetry.redaction.failure and android.telemetry.export.status.
• CLI output from the injector/status helpers and any queue export logs.
• The override log entry plus a link to the relevant section in android_runbook.md and the chaos rehearsal transcript.

Incidents without the artefacts above remain action items during the next SRE governance review.

How do salt rotations work on Android?

• Salt epochs are governed by the policy in telemetry_redaction.md.
• During the pre-flight checks, compare the reported android.telemetry.redaction.salt_version against the quarterly secret stored in the vault (scripts/telemetry/check_redaction_status.py --status-url …).
• When rotating, run the same command with --write-cache to capture the before and after timestamps, then store the diff under docs/source/sdk/android/readiness/screenshots/<YYYY-MM-DD>/salts/.
• Rotations must be announced in #android-program and referenced in the override audit log if an incident overlaps the change.

How do we verify parity with the Rust telemetry schema?

Use scripts/telemetry/run_schema_diff.sh with the Android and Rust schema snapshots (configs/android/telemetry_schema.json and configs/rust/telemetry_schema.json). The output feeds both telemetry_schema_diff.md and the FAQ entry for the relevant release. Any new field requires:

1. Update to the policy doc (telemetry_redaction.md, signal inventory table).
2. An alert or dashboard note describing the retention/redaction plan.
3. Inclusion in the enablement packet and knowledge check questions.

What is the chaos rehearsal workflow?

The rehearsal plan lives in telemetry_chaos_checklist.md. Each scenario must:

• Execute the checklist commands (injector, exporter, queue snapshot, recovery).
• Capture logs via ci/run_android_telemetry_chaos_prep.sh --status-only.
• Archive screenshots and CLI transcripts under docs/source/sdk/android/readiness/screenshots/<scenario>/.
• Produce an FAQ update summarising lessons learned and any new mitigation steps. Link back to the checklist item so future sessions stay deterministic.

Where do training assets and knowledge checks live?

• Pre-read: telemetry_redaction_pre_read.md
• Workshop minutes: telemetry_redaction_minutes_20260212.md
• Quick reference: telemetry_redaction_quick_reference.md
• Chaos lab artefacts: docs/source/sdk/android/readiness/
• Knowledge check + FAQ updates: This file

After each enablement session, append the new questions/answers here and update the knowledge-check forms referenced in telemetry_redaction_minutes_20260212.md.

How do I propose a new FAQ entry?

1. Draft the question/answer, including links to supporting evidence (runbooks, scripts, screenshots).
2. Update this file and add the corresponding reference to status.md if the change modifies readiness evidence.
3. Post a short summary in #and7-telemetry so the Android Observability TL can track the backlog.

Future sessions can tag unresolved topics with “🈸 Drafting” to keep parity with the roadmap status legend.