Skip to content

chore(deployment): clarify TDX HW requirements are per-CVM, bump disk to 1TB#3449

Merged
netrome merged 3 commits into
mainfrom
docs/tdx-host-hw-requirements
Jun 3, 2026
Merged

chore(deployment): clarify TDX HW requirements are per-CVM, bump disk to 1TB#3449
netrome merged 3 commits into
mainfrom
docs/tdx-host-hw-requirements

Conversation

@barakeinav1

@barakeinav1 barakeinav1 commented Jun 3, 2026

Copy link
Copy Markdown
Contributor

Closes #3448

The listed 64GB/8-core/500GB specs are what a single CVM consumes, not
the TDX host. Operators may run two CVMs concurrently during a launcher
migration, so size the host for ~2x. Also bump the CVM disk to 1TB since
500GB is too small for a mainnet node.

Closes #3448

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Copilot AI review requested due to automatic review settings June 3, 2026 04:54
@claude

claude Bot commented Jun 3, 2026

Copy link
Copy Markdown

Pull request overview

Clarifies that the TDX hardware requirements listed in the external operator guide are per-CVM, since operators may run two CVMs concurrently during a launcher migration. Also bumps the documented CVM disk from 500GB to 1TB (1000GB), which the PR description notes is too small for mainnet. Doc-only change, no code touched.

Changes:

  • Adds a callout that the listed figures are per-CVM and operators should size the host for ~2x.
  • Appends "per CVM" to the memory, vCore, and disk bullets.
  • Updates the Dstack web-interface deployment instructions to use disk = 1000 GB.

Reviewed changes

Per-file summary
File Description
docs/running-an-mpc-node-in-tdx-external-guide.md Adds per-CVM clarification + 2x sizing note in Hardware Requirements; bumps disk to 1TB in both the requirements list and the Dstack web-UI VM settings step.

Findings

Non-blocking (nits, follow-ups, suggestions):

  • localnet/tee/scripts/rust-launcher/set-testnet-env.sh:28, localnet/tee/scripts/rust-launcher/set-localnet-env.sh:9, localnet/tee/scripts/rust-launcher/deploy-tee-cluster.sh:56, localnet/tee/scripts/rust-launcher/how-to-run-deploy-tee-cluster.md:131 — these comments still cite 500 GB disk per CVM. They're localnet/testnet sizing notes so 500GB may still be adequate there, but if the 1TB figure reflects the true per-CVM disk footprint (not just a mainnet-only floor), these should be updated to keep operator-facing guidance consistent. Out of scope for this PR if intentional — flagging so it isn't lost.

✅ Approved

Copilot AI 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.

Pull request overview

Updates the external TDX deployment guide to clarify that the stated compute/memory/storage requirements are per Confidential VM (CVM), and adjusts disk sizing guidance for mainnet operators who may run two CVMs concurrently during launcher migrations.

Changes:

  • Clarify that resource figures are per-CVM and advise sizing the TDX host for ~2x during migrations.
  • Increase the recommended CVM disk size from 500GB to ~1TB and update the VM HW settings accordingly.

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

Comment thread docs/running-an-mpc-node-in-tdx-external-guide.md Outdated
Comment thread docs/running-an-mpc-node-in-tdx-external-guide.md Outdated
barakeinav1 and others added 2 commits June 3, 2026 05:05
Address review feedback: the per-CVM "2x the host" note no longer
covers the host-level CPU/RAM-slot requirement, and the disk figure
reads "1TB (1000 GB)" to match the VM setting and max_disk_size config.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Align the deployment configs with the 1TB per-CVM guidance: the
cvm-deployment templates (default.env, sgx/kms configs,
deploy-launcher-guide.md) and the testnet node envs (frodo, sam) all
defaulted to DISK=500G. Localnet/test scaffolding left at 500G/128G.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
@claude

claude Bot commented Jun 3, 2026

Copy link
Copy Markdown

PR title type suggestion: This PR changes both documentation and deployment config files (.env), so the type prefix should probably be chore: instead of docs:.

Suggested title: chore: clarify TDX HW requirements are per-CVM, bump disk to 1TB

@barakeinav1

barakeinav1 commented Jun 3, 2026

Copy link
Copy Markdown
Contributor Author

addressed the production side in 589eac5: bumped DISK=500G1000G in the deployment/cvm-deployment/ configs (default.env, sgx/kms, deploy-launcher-guide.md) and the testnet node envs (frodo, sam). Left the localnet/test 500G/128G, since those sizes are adequate for local/CI runs.

@barakeinav1 barakeinav1 changed the title docs: clarify TDX HW requirements are per-CVM, bump disk to 1TB chore(deployment): clarify TDX HW requirements are per-CVM, bump disk to 1TB Jun 3, 2026

@netrome netrome left a comment

Copy link
Copy Markdown
Collaborator

Choose a reason for hiding this comment

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

Thanks!

@netrome netrome merged commit 2d9d583 into main Jun 3, 2026
16 checks passed
@netrome netrome deleted the docs/tdx-host-hw-requirements branch June 3, 2026 18:30
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.

Docs: TDX hardware requirements list per-CVM specs, not host specs

3 participants