Imported from jeremylongshore/tons-of-skills-marketplace (skills/.curated/coreweave-upgrade-migration/SKILL.md). Install upstream with npx skills add jeremylongshore/tons-of-skills-marketplace --skill coreweave-upgrade-migration. Copyright stays with the author (MIT).

CoreWeave Upgrade & Migration

Community-contributed. Not affiliated with, endorsed by, or sponsored by CoreWeave, Inc. CoreWeave is a registered trademark of CoreWeave, Inc.

Overview

CoreWeave is a GPU-specialized cloud provider running Kubernetes-native infrastructure. Migrations involve upgrading between GPU instance types (A100 to H100), updating CUDA driver versions, and handling Kubernetes API version changes across namespaces. Tracking API versions is critical because CoreWeave's instance type labels and resource quotas change between platform releases, and deploying to a deprecated instance class will cause scheduling failures.

Version Detection

import { KubeConfig, CoreV1Api } from "@kubernetes/client-node";

async function detectCoreWeaveVersion(): Promise<void> {
  const kc = new KubeConfig();
  kc.loadFromDefault();
  const k8sApi = kc.makeApiClient(CoreV1Api);

  // Check current namespace GPU allocations
  const pods = await k8sApi.listNamespacedPod("my-namespace");
  for (const pod of pods.body.items) {
    const gpuClass = pod.spec?.nodeSelector?.["gpu.nvidia.com/class"];
    const cudaVersion = pod.metadata?.labels?.["cuda-version"];
    console.log(`Pod ${pod.metadata?.name}: GPU=${gpuClass}, CUDA=${cudaVersion}`);
  }

  // Detect deprecated instance types
  const deprecated = ["A100_PCIE_40GB", "V100_PCIE_16GB", "RTX_A5000"];
  const activeGpus = pods.body.items
    .map((p) => p.spec?.nodeSelector?.["gpu.nvidia.com/class"])
    .filter(Boolean);
  const stale = activeGpus.filter((g) => deprecated.includes(g!));
  if (stale.length > 0) console.warn(`Deprecated GPU types in use: ${stale.join(", ")}`);
}

Migration Checklist

• Review CoreWeave release notes for deprecated instance types
• Audit all deployments for gpu.nvidia.com/class node selectors
• Verify CUDA version compatibility with target GPU (see matrix below)
• Update container base images to match new CUDA/cuDNN requirements
• Test inference latency on new GPU type in staging namespace
• Update resource requests (nvidia.com/gpu) for new instance memory
• Migrate persistent volumes if switching regions or availability zones
• Update Kubernetes API version in manifests (e.g., apps/v1 changes)
• Validate HPA scaling behavior on new instance type throughput
• Run canary deployment with traffic split before full cutover

Schema Migration

// CoreWeave instance type labels changed in 2025 platform update
// Old: gpu.nvidia.com/class: "A100_PCIE_80GB"
// New: gpu.nvidia.com/class: "H100_SXM5_80GB"

interface DeploymentMigration {
  oldSelector: Record<string, string>;
  newSelector: Record<string, string>;
  cudaMinVersion: string;
}

const GPU_MIGRATIONS: DeploymentMigration[] = [
  {
    oldSelector: { "gpu.nvidia.com/class": "A100_PCIE_80GB" },
    newSelector: { "gpu.nvidia.com/class": "H100_SXM5_80GB" },
    cudaMinVersion: "12.4",
  },
  {
    oldSelector: { "gpu.nvidia.com/class": "A100_SXM4_80GB" },
    newSelector: { "gpu.nvidia.com/class": "H100_SXM5_80GB" },
    cudaMinVersion: "12.4",
  },
];

function migrateNodeSelector(manifest: any, migration: DeploymentMigration): any {
  const selector = manifest.spec?.template?.spec?.nodeSelector;
  if (!selector) return manifest;
  for (const [key, oldVal] of Object.entries(migration.oldSelector)) {
    if (selector[key] === oldVal) {
      selector[key] = migration.newSelector[key];
    }
  }
  return manifest;
}

Rollback Strategy

Prerequisites

• A versioned manifest, compatible CUDA/image matrix, and capacity confirmation for the target GPU or cluster change.
• Staging evaluation evidence, named migration owner, and a verified previous deployment revision.
• Backup/retention approval for any PVC, checkpoint, or model-artifact move.

Instructions

1. Inventory current selectors, images, API versions, volumes, quotas, and service SLOs.
2. Update the manifest in staging, validate server-side, and run compatibility and performance checks.
3. Promote through a bounded canary only after the readiness, quality, latency, and integrity gates pass.
4. Use the prior immutable revision on any failed gate; do not delete old data or nodes until the observation period closes.

import { AppsV1Api, KubeConfig } from "@kubernetes/client-node";

async function rollbackDeployment(namespace: string, name: string): Promise<void> {
  const kc = new KubeConfig();
  kc.loadFromDefault();
  const appsApi = kc.makeApiClient(AppsV1Api);

  // Kubernetes rollout undo — reverts to previous revision
  const deployment = await appsApi.readNamespacedDeployment(name, namespace);
  const currentRevision = deployment.body.metadata?.annotations?.["deployment.kubernetes.io/revision"];
  console.log(`Rolling back ${name} from revision ${currentRevision}`);

  // Patch to trigger rollback via revision annotation
  await appsApi.patchNamespacedDeployment(name, namespace, {
    spec: { template: { metadata: { annotations: { "kubectl.kubernetes.io/restartedAt": new Date().toISOString() } } } },
  }, undefined, undefined, undefined, undefined, undefined, { headers: { "Content-Type": "application/strategic-merge-patch+json" } });
  console.log(`Rollback initiated for ${name} in ${namespace}`);
}

Error Handling

Output

• A reviewed migration plan with compatibility results, capacity decision, owner, and rollback revision.
• A staged/canary receipt proving readiness, SLO, and artifact-integrity checks.
• A reversible fallback that preserves the old workload and approved data until sign-off.

Examples

Test the target manifest in staging before moving a production selector:

kubectl -n inference-staging apply --dry-run=server -f migration.yaml
kubectl -n inference-staging apply -f migration.yaml
kubectl -n inference-staging rollout status deployment/inference --timeout=15m

If CUDA compatibility, scheduling, or the evaluation gate fails, undo the staging revision and stop promotion. Do not change production affinity or delete the old PVC to make a migration appear complete.

Resources

• CoreWeave GPU Instances
• CoreWeave Kubernetes Docs
• CoreWeave Changelog

Next Steps

For CI/CD pipeline integration, see coreweave-ci-integration.