01 // OVERVIEW
CHECKING
Pod & Hardware Summary
ACCELERATOR ARCHITECTURE
NVIDIA A40
48 GB GDDR6 ECC • 1x GPU • CUDA 13.2
COMPUTE AVAILABILITY
—
- Pod Name
—- Pod ID
—- Lifecycle State
—- Datacenter / Tier
—- Host Allocation
—- Session Uptime
—- Last Started (UTC)
—- Persistent Volume
/workspace (GeeseFS)
02 // LIFECYCLE CONTROLS
READY
Power & State Management
Execute verified state transitions via RunPod REST API v2
(POST /v2/pods/{id}/action). Actions are locked to the configured Pod and
polled until convergence is confirmed.
Checking allowed lifecycle transitions...
Active Control Plane Safeguards
- Duplicate start/stop prevention & server-side concurrency lock
- Bounded state convergence polling before reporting completion
- SSH telemetry isolation (SSH faults never block API lifecycle controls)
03 // LIVE GPU TELEMETRY
Real-Time NVIDIA A40 Metrics (SSH
Real-Time NVIDIA A40 Metrics (SSH nvidia-smi)
SSH: CHECKING
GPU Core Utilization
—
Memory Controller:
—
P-State: —
VRAM Used / Total
—
Utilization:
—
Free: —
GPU Die Temperature
—
Fan Speed:
—
Thermal Limit: 90°C
Power Consumption
—
Envelope:
—
Clocks: —
Hardware & Runtime Telemetry
- GPU UUID
—- NVIDIA Driver
—- Remote System RAM
—- Container Overlay (
/) —- Persistent Mount (
/workspace) —- Load Average (1m / 5m / 15m)
—
Active GPU Compute Processes
No active GPU compute processes reported.
04 // COST & BILLING
ESTIMATE
Compute Rate & Estimated Spend
Configured GPU Hourly Rate
—
RunPod API reported rate
Current Active Burn Rate
—
Based on current Pod state
Current Session Compute (Est.)
—
Uptime × hourly rate (Estimate)
Projected Active Compute Cost (Clearly Labelled Estimates)
1 Hour Active
—
4 Hours Active
—
8 Hours Active
—
24 Hours Active
—
Active Compute Charges:
While the Pod status is
RUNNING, RunPod bills the NVIDIA A40 compute rate per second.
Stopped Pod Storage vs. Compute Charges:
When the Pod is stopped (
EXITED), GPU compute billing drops to $0.00/hr,
but RunPod continues charging for retained container disk and persistent /workspace
storage. Because RunPod REST API v2 does not expose a separate stopped-storage rate on the Pod
endpoint, no storage dollar figure is invented here.
05 // CONNECTION HEALTH
CHECKING
Provider API & SSH Transport Diagnostics
RunPod REST API v2
https://api.runpod.io/v2
- HTTP Status
—- Round-Trip Latency
—- Last Successful Check
—- Diagnostic Status
Nominal
SSH Telemetry Channel (
runpod-ai-avatar)—
- Host Key Verification
Strict (known_hosts)- Probe Latency
—- Last Successful Check
—- Diagnostic Status
—