Skip to content

Ranges

Launch and manage isolated demo environments.

Range Lifecycle

stateDiagram-v2
    [*] --> Pending: Launch
    Pending --> Provisioning: Processing
    Provisioning --> Ready: Complete
    Provisioning --> Failed: Error
    Ready --> Destroyed: Destroy
    Pending --> Destroyed: Cancel
    Provisioning --> Destroyed: Cancel
    Failed --> Destroyed: Destroy

Status Reference

Status Meaning
Pending Queued for provisioning
Provisioning Infrastructure being created
Ready Range is live, accessible
Failed Provisioning error occurred
Destroyed Range terminated

Launch a Range

  1. Go to Ranges page
  2. Select a scenario
  3. Select an agent
  4. Click Launch Range

Provisioning takes about 10 minutes.

Monitor Provisioning

The Ranges page shows real-time status updates during provisioning. You'll see progress as instances are created and configured.

Access a Range

Once Ready:

  1. Go to Terminal
  2. Select an instance
  3. Use SSH or RDP to connect

When the active range advertises VPN access, Mission Control also offers a Download VPN profile action. Treat the .ovpn file as a private credential.

See Terminal for details.

Range Lifetime

Mission Control ranges have a server-owned lease: an initial lifetime, a per-extension increment, and a maximum lifetime a range can never exceed. The SPA shows the time remaining and the scheduled cleanup time on the active range page, and offers Extend by up to N days (the configured increment), bounded by the range's maximum lifetime. Expired ranges are automatically destroyed.

The default lease is 30 / 30 / 365 days (initial / increment / maximum).

CTF participant and spare ranges use the CTF event cleanup time as their deadline and are removed automatically through the same range cleanup process; they are unaffected by the Mission Control lease policy below.

Configuring the lease policy

The deployment setting is the fallback policy. Set it in bigrae.yaml under settings.mission_control_leases; the same provider-neutral block applies to both AWS and GCP:

settings:
  mission_control_leases:
    initial_days: 30        # lifetime a new range starts with (positive)
    extension_days: 30      # increment each extension adds (positive)
    maximum_days: 365       # ceiling a range can never exceed (positive, >= initial_days)
    extensions_enabled: true

Rules and behavior:

  • All fields are optional; omitted fields use the defaults above, so an omitted block keeps the historical 30 / 30 / 365 behavior.
  • initial_days must be positive and no greater than maximum_days; extension_days must be positive. extension_days may exceed maximum_days; an extension is always bounded by the range's remaining lifetime. Invalid combinations fail configuration validation at install time, not at range launch.
  • New ranges only. Durations are snapshotted onto each range generation when it is created. Changing the policy applies to ranges created afterward; ranges that already exist keep the deadlines and increment they launched with. Lowering maximum_days does not shorten existing ranges, and raising it does not extend their ceilings.
  • extensions_enabled: false is a live switch that denies further extensions across all Mission Control ranges after the deployment rolls out. It does not shorten any deadline and does not stop automatic cleanup; expired ranges are still destroyed. Use it to stop new extensions without orphaning ranges. Setting initial_days equal to maximum_days only prevents extensions for new ranges and is not a substitute for the switch.
  • Cost. Longer leases increase standing infrastructure, storage, and access exposure for the lease duration; shorter leases reduce cost but can expire while a range is still in use. Cleanup is asynchronous destruction dispatch, so billing stops when destruction completes, not exactly at the deadline.

Deployment-fallback changes take effect after the platform processes roll out with the new configuration (the runtime ConfigMap change triggers a normal rolling restart); mixed old/new processes may briefly apply different fallback policies until the rollout completes.

Runtime tenant and group policy (platform administrators)

An active platform superuser can change the effective Mission Control policy at Administer → Platform settings without redeploying BigRAE. The page always shows the deployment fallback and whether the tenant currently uses that fallback or a runtime replacement.

  • Saving the tenant policy replaces all four fields as one revision-checked operation. Reset tenant policy removes the runtime row and restores the current deployment fallback.
  • A policy on an administrator-controlled Django RBAC group replaces the tenant values for that group. If a user belongs to several configured groups, the lowest initial, extension, and maximum durations win and extensions are enabled only when every matching group and the tenant allow them. Group initial and maximum durations cannot exceed the effective tenant maximum.
  • CTF Participant, provider claim strings, CTF teams, and workspace or organization roles never select a Mission Control policy. Group policy restricts lease behavior; it does not grant launch or range access.
  • Runtime duration changes affect new cold-created and warm-claimed generations only. Existing generations retain their snapshotted deadlines and increment. The current tenant/group extensions_enabled switch is checked live and can stop another extension without shortening a range or disabling cleanup.
  • Each save or reset is strict-audited and uses the revision displayed by the page. A stale browser receives a conflict and must refresh instead of silently overwriting a newer change. Reset advances that revision even though the override is removed, so a later replacement cannot reuse an old revision.

Cancel a Range

While in Pending or Provisioning status:

  1. Go to Ranges page
  2. Click Cancel on the range

Destroy a Range

When finished:

  1. Go to Ranges page
  2. Click Destroy on the range

This is irreversible. All range data is deleted.

Warm pool for faster initial launch

A deployment can keep a warm pool of pre-provisioned, system-owned ranges so an initial launch is handed a ready range instead of waiting for cold provisioning. Warm pooling is disabled by default; an operator enables it in bigrae.yaml under settings.warm_pool. An active pool also requires WARM_POOL_SERVICE_CREDENTIAL_UUID, set to the UUID of an active, installation-scoped service admission with installation.launch_range. The UUID is a non-secret reference; the workload still authenticates through its normal keyless service identity. Startup fails if an active policy has no valid reference.

How it behaves:

  • Launch first tries to atomically claim a compatible ready range from the pool. On a hit, the range is transferred to the launching principal and account scope with fresh, fully re-established credentials and access. No prior tenant's data, credentials, or sessions cross the ownership boundary. On a miss (empty pool, an incompatible request, or a backend without warm support), launch falls back to normal cold provisioning with no change in behavior.
  • Compatibility is decided by the exact immutable launch inputs (backend, region, scenario package/lock digest, purpose, access posture). A scenario, image, or config change makes older ready ranges incompatible; the pool retires them and prepares new ones.
  • Supported backends: the GCE range-cell backend is warm-capable today. AWS and the retained GDC substrate report warm activation as unsupported and always cold-provision (AWS support is tracked separately).

Sizing and cost:

  • Each bucket declares a target, minimum, and maximum ready count and a warm idle lifetime. Warm ranges count against provider capacity and cost admission exactly like a launched range. A warm pool trades standing cost for lower launch latency, so size target to your expected concurrent cold-start demand, not higher.
  • A capacity or cost ceiling can hold the pool below minimum; the reconciler alerts rather than exceeding a ceiling.
  • Metrics are published to the Shifter/WarmPool namespace: ready / provisioning / unhealthy counts, claim hit and fallback rates, and idle age, per bucket and backend/region.

This warm pool is distinct from the CTF recovery-spare pool (see the CTF organizer guide): recovery spares replace a failed participant range mid-event, while the warm pool speeds up initial launches. The two are configured and accounted separately.

Limits

  • One active range at a time per user
  • Mission Control ranges cannot be extended past their maximum lifetime (default 365 days from creation; see Configuring the lease policy)
  • Warm-pool claims apply to RAES-native (GCE) initial launches; other backends cold-provision.