Rack Settings
The Rack Settings page in the Console provides configuration controls for Rack-wide features. Navigate to it from the sidebar by selecting a Rack and clicking Settings.
Role Badge
A badge in the Rack header shows your effective role on the current Rack:
| Badge | Access Level |
|---|---|
| Read-only | View Rack, Apps, and settings. Cannot modify. |
| Read+Write | Full operational access (deploy, scale, configure). |
| Administrator | All permissions including member management and destructive actions. |
The badge links to the role legend page for your organization. Pre-3.24.6 Racks do not display a role badge.
Cost Tracking
The Cost Tracking card shows whether per-App spend accumulation is enabled on the Rack. Cost tracking powers budget caps, alerts, and auto-shutdown features.
Current status is displayed as an Enabled or Disabled badge.
To toggle, run the CLI command shown on the card:
$ convox rack params set cost_tracking_enable=true -r <rack-name>
Cost tracking requires Rack version 3.24.6 or later. The card displays a version gate message on older Racks.
See Cost Tracking for setup and usage details.
GPU Telemetry Scraper
The GPU Telemetry Scraper card controls the Prometheus scraper that collects GPU metrics from DCGM exporters on each node.
Prerequisites:
- The DCGM Exporter (
gpu_observability_enable) must be enabled first. The scraper toggle is locked until DCGM is active. - Rack version 3.24.6 or later.
Toggle the scraper on or off directly from the card. The toggle responds immediately and stays interactive during transitions. Status states:
| Status | Meaning |
|---|---|
| Enabled | Scraper is running and collecting metrics |
| Enabling... | Rack update in progress to install the scraper |
| Disabled | Scraper is not installed |
| Disabling... | Rack update in progress to remove the scraper |
| Error: ... | Worker failure; admin reset may be available |
Admin Reset: When the scraper enters a frozen error state, administrators see a Reset link to clear it. The toggle remains interactive so you can request a new state regardless of the current error.
Toggling applies a non-disruptive Rack update. Running workloads are not affected.
See GPU Metrics for Rack-side setup and GPU Dashboard for the Console monitoring view.
KEDA Status
The KEDA card shows whether Kubernetes Event-Driven Autoscaling is enabled on the Rack. KEDA unlocks GPU utilization, inference queue depth, custom Prometheus, and message-queue autoscale triggers. CPU and memory autoscaling works without KEDA.
To toggle, run the CLI command shown on the card:
$ convox rack params set keda_enable=true -r <rack-name>
The card includes expandable examples for:
- Classic autoscale (CPU/memory targets, no KEDA required)
- KEDA-based autoscale (GPU, queue depth, CPU thresholds)
- Direct KEDA scaler (raw KEDA trigger config in convox.yml)
See KEDA Autoscaling for full documentation.
Webhook Signing Key Management
The Webhook Signing Key section manages the HMAC-SHA256 key used to sign outbound webhooks. Receivers verify the signature header (Convox-Signature: t=...,v1=...).
Actions (Administrator only):
- Generate first key: Creates the initial signing key when none exists.
- Reveal: Shows the current key value (auto-masked after display).
- Rotate: Generates a new key. Up to 4 keys remain active during rotation so receivers can transition. The oldest key is evicted when the limit is reached.
A rotation-in-progress banner appears when multiple keys are active, prompting you to update receivers and confirm the rotation.
Non-administrator users see a message indicating that administrator access is required.
See Webhook Signing for receiver verification instructions.
Rack Uninstall
The Uninstall Rack card is a destructive action that permanently removes all Apps, resources, and infrastructure associated with the Rack. This action cannot be undone.
Click Uninstall Rack to open a confirmation dialog. The uninstall process tears down all Terraform-managed infrastructure for the Rack.
See Also
- CLI Rack Management for managing Racks from the command line
- Console Rack Management for moving Racks between CLI and Console
- RBAC for configuring organization access controls
- Autoscale Triggers for Console-driven autoscaling
- GPU Dashboard for GPU telemetry visualization
- Budget Management for per-App spend tracking and caps
