Skip to content

docs: add guides for migrating to Runpod from 14 other platforms - #968

Open
lukepiette wants to merge 3 commits into
mainfrom
docs/migration-guides
Open

lukepiette wants to merge 3 commits into
mainfrom
docs/migration-guides

Conversation

@lukepiette

Copy link
Copy Markdown
Contributor

What

Adds a "Migrate to Runpod" section to the docs: an overview page and one end-to-end guide per platform.

  • GPU clouds: Vast.ai, Lambda, CoreWeave, Paperspace and DigitalOcean
  • Serverless and inference platforms: Modal, Replicate, Baseten, fal, Together AI, Fireworks AI, Hugging Face
  • Hyperscalers: AWS, Google Cloud, Azure (each with a Terraform section using the 2.x provider)

Every guide has the same shape: a concept map, what works differently and what has no direct equivalent (in both directions), eight steps from inventory to shutdown with before and after code, reference maps (CLI, API, config fields, env vars, statuses), a prompt for coding agents, and troubleshooting.

Other changes:

  • docs.json: new "Migrate to Runpod" nav group above Reference. The existing Cog and OpenAI migration pages move into it (URLs unchanged, so no redirects).
  • get-started/agent-skills.mdx: lists the runpod-cloud-migrate skill and the /runpod:migrate-cloud command.

Covers DR-1650 (Vast.ai guide).

How it was checked

  • Each guide was written against the other platform's current official docs and against these docs and the REST v2 OpenAPI spec, then fact-checked by a second, independent pass that made about 120 corrections.
  • Every fenced Python, Bash, JSON and YAML block parses. HCL, JavaScript and Dockerfiles were read by eye.
  • mint broken-links passes.
  • Rules applied throughout: no prices or price comparisons for other platforms, no availability promises, no GraphQL or REST v1 recommendations, and only runpodctl commands that are in the CLI reference here.

Not done, and worth a reviewer's attention

  • Nothing was run live. No Pod, endpoint, volume or terraform apply was created, and no before/after program was executed. A live pass on the Runpod half of a few guides (Vast.ai, Modal, AWS, one vLLM base URL swap) is the highest-value check before publishing.
  • ECR on Serverless: the REST v2 spec says registry delegations apply to Serverless, but tutorials/pods/use-private-ecr-images.mdx says Serverless uses a different mechanism. The AWS guide tells readers to test first.
  • Stated gaps are public. Each guide lists what Runpod doesn't have versus that platform (for example UDP, spot Pods through REST v2, VPCs, cron, signed webhooks). Product should confirm the wording.
  • Other platforms change fast. Their facts were current on 2026-11-14.
  • The pages weren't viewed in a rendered Mintlify preview.

Depends on

runpod/runpod-plugins-official#127 adds the runpod-cloud-migrate skill and /runpod:migrate-cloud command that every guide's "Migrate with a coding agent" section points to. That should merge and release before this publishes.

🤖 Generated with Claude Code

Add a "Migrate to Runpod" section: an overview page and a guide each for
Vast.ai, Lambda, CoreWeave, Paperspace and DigitalOcean, Modal, Replicate,
Baseten, fal, Together AI, Fireworks AI, Hugging Face, AWS, Google Cloud
and Azure. The existing Cog and OpenAI migration pages move into the new
nav group, and the agent skills page lists the runpod-cloud-migrate skill.

Co-Authored-By: Claude Fable 5.2 <noreply@anthropic.com>
@mintlify

mintlify Bot commented Nov 14, 2026 •

Copy link
Copy Markdown
Contributor

Preview deployment for your docs. Learn more about Mintlify Previews.

Project Status Preview Updated
runpod-docs 🟢 Ready View Preview Nov 14, 2026, 8:07 PM

💡 Tip: Enable Automations to automatically generate PRs for you.

A mechanical pass validated every runpodctl invocation, REST v2 body, SDK
symbol and Terraform block against the command surface, the OpenAPI spec, the
runpod package source and the provider schemas. Four adversarial passes
re-checked the other platforms' facts against their live docs and walked each
migration end to end.

Data-loss and billing fixes:
- CoreWeave: shutdown no longer deletes the PVC with the Deployment
- Lambda: copy root-volume data before terminating instances
- Google Cloud: warn that terraform import can force Pod replacement
- Vast.ai: shutdown now deletes Deployments; rsync step needs a key the Pod accepts
- Baseten: orchestrator polls a long job instead of resubmitting it

Runpod-side corrections:
- vLLM worker settings the current worker ignores (structured output,
  speculative decoding), and routes it does serve (Responses, Messages)
- Remove Public Endpoints that are not in the live catalog
- Custom handlers point at the cached-model path so weights are not downloaded on each cold start
- Load-balancing endpoints: set ports, and a timeout within the request limit
- serverless update scaling flags, zero-balance behavior, global networking scope

Co-Authored-By: Claude Fable 5.2 <noreply@anthropic.com>
runpodctl v3.1.1 lists config as deprecated. Checked all 804 runpodctl
invocations in the guides and playbooks against a v3.1.1 build: these two
rows were the only ones using a deprecated command.

Co-Authored-By: Claude Fable 5.2 <noreply@anthropic.com>

@nick4fake nick4fake left a comment

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I think this should go to KB, not docs. If it stays here:

  • hold the merge until runpod-plugins-official#127 is merged and released, all pages point to runpod-cloud-migrate / /runpod:migrate-cloud
  • codex isn't a valid --client for npx @runpod/mcp-server add (overview, vast-ai and aws tabs), Codex gets MCP from the plugin. Maybe move this block to a snippet
  • aws, SageMaker on LB endpoint: serve.py keeps the model and input on CPU, needs map_location="cuda" and the tensor on cuda like inference.py
  • aws step 5: A4000 just to run aws s3 sync, a CPU pod is enough (--compute-type cpu)
  • ECR on serverless: I wouldn't publish "test first", let's check which source is right and fix it
  • lots of numbers copied from other pages (spend limit, LB limits, payload sizes etc), better to link them

I only read overview, vast-ai, modal and aws.

Comment thread migrate/overview.mdx
npx @runpod/mcp-server@latest add --client CLIENT_ID
```

Replace `AGENT_ID` and `CLIENT_ID` with your agent's ID, for example `cursor` or `codex`.

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

codex isn't a --client id, cursor or opencode instead

This branch was successfully deployed

1 active deployment
staging — 5284e02a Deployed Nov 14, 2026 by mintlify[bot]
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

3 participants