Data, retention & support

The Agent sends operational metadata such as host identity, device identifiers, pool and dataset names, capacity, performance counters, maintenance state, and deployment posture. Observer adds aggregated I/O distributions. The documented deployment excludes media dataset mounts and does not collect media file contents.

Metadata can still identify a customer or production. Avoid confidential project names in host or dataset labels where your policy prohibits transmitting them. Metadata redaction only masks selected identifiers; it does not anonymize the entire payload. Review Security and the Privacy Notice, including account and website data.

Configured alert destinations receive incident details. Only route them to approved services and recipients, and check their access and retention policies.

The current pruning policy keeps disk telemetry up to 30 days for Boutique and trial/default accounts, 90 days for Studio, and 365 days for Facility. Other plan identifiers should be confirmed with support rather than inferred from their display name. Retention describes cleanup windows, not a guarantee of continuous or complete observations.

Observer raw latency windows are kept for 7 days and 1-minute rollups for 30 days. Coarser latency rollups have 90-day and 365-day ceilings with plan-specific pruning. Host network samples use separate windows: 14 days raw, 90 days at 5-minute resolution, 365 days hourly, and 400 days for link events. Posture-change history has a 365-day ceiling.

Longer retention does not mean raw resolution for the whole period. A chart or export may expose a shorter window. Incident evidence, account information, and billing records follow separate lifecycles; telemetry pruning is not an account-deletion process. Export needed reports before offboarding.

1. Check telemetry freshness, the active organization, host connectivity, and service status.
2. Consult Troubleshooting for preflight codes and connectivity checks.
3. Use the support contact in your service agreement. For account or policy contact details, see the Terms. Include the affected page, approximate time and timezone, component versions, platform, error text, and steps to reproduce.
4. Redact claim tokens, API keys, webhook URLs, credentials, and confidential host or dataset names from logs and screenshots. Do not send media files.

Support availability and response commitments depend on your agreement.

1. Export the operational and billing evidence you need to retain.
2. Revoke the node key in Settings → Agent Pairing, then disable updater/startup jobs and remove Agent and Observer components using the uninstallation guide.
3. Remove unneeded alert destinations and organization memberships. Manage subscription cancellation separately through billing.
4. For account or personal-data requests, follow the Privacy Notice. Uninstalling a container does not itself erase cloud history.