Backup and restore

A reproducible research platform needs recoverable metadata, workspace data, and telemetry. Backup ownership depends on the deployment model.

What to back up

Data	Why it matters
PostgreSQL	Projects, identity mappings, operations, resource metadata.
Object storage	Workspace data, event payloads, and export bundles.
Event store	Queryable telemetry.
Secrets	API signing keys, service credentials, customer-managed references.
Helm values	Deployment configuration.
Container image digests	Rebuild and rollback evidence.
Audit logs	Security and compliance evidence.

Priority

Priority	Data
P0	Identity/session keys, PostgreSQL, object storage, event store.
P1	Helm values, image digests, registry mirrors, audit logs.
P2	Derived caches and rebuildable runtime state.

Restore checklist

Restore database to a known point.
Restore object storage required by workspace data and event payloads using an exact per-object version manifest whose entries are at or before the selected database restore point.
Restore or rotate signing and service credentials.
Reapply Helm values for the target environment.
Verify API identity with hiloop whoami.
Verify telemetry query for a known run.
Verify a known workspace object and its content digest.
Verify a bounded sandbox create, exec, and delete.

For hosted deployments, backup and restore responsibilities are handled by the managed service. For BYOC, hybrid, and on-prem deployments, document the exact ownership and recovery targets during deployment planning.

​What to back up

​Priority

​Restore checklist

What to back up

Priority

Restore checklist