Pods (Containers)

Pods are containerized workloads that run on Podstack’s Kubernetes infrastructure. They provide a fast, flexible way to deploy GPU-accelerated applications.

What is a Pod?

A pod is one or more containers running together with shared resources. On Podstack, pods typically run a single container with:

  • GPU access (optional)
  • CPU and memory allocation
  • Storage mounts
  • Network connectivity
  • SSH and web terminal access

Key Features

GPU Support

  • Whole GPU allocation (1, 2, 4, or more GPUs)
  • Multiple GPU types (A100, H100, V100, L40S, T4)
  • CUDA and cuDNN pre-installed in most images

Container Images

Use any Docker image:

  • Public images from Docker Hub, NGC, etc.
  • Private registry images with authentication
  • Custom images built for your workload

Access Methods

  • SSH: Direct terminal access via assigned subdomain
  • Web Terminal: Browser-based terminal
  • Jupyter Notebook: Built-in notebook server (if enabled)
  • Custom Ports: Expose any ports for web services

Volume Mounts

  • Mount NFS volumes for persistent shared storage
  • ConfigMaps for configuration files
  • SSH keys automatically mounted for access

Pod Lifecycle

Creating → Pending → Running → (Stopped) → Terminated
StateDescriptionBilling
CreatingPod being provisionedNo
PendingWaiting for resourcesNo
RunningPod is activeYes
StoppedPaused by userNo
TerminatedPod deletedNo

Tip: Stop pods when not in use to pause billing while preserving configuration.

In This Section