Skip to content

docs: add KMS upgrade plan for operators#705

Draft
Leechael wants to merge 2 commits into
masterfrom
docs/kms-upgrade-plan
Draft

docs: add KMS upgrade plan for operators#705
Leechael wants to merge 2 commits into
masterfrom
docs/kms-upgrade-plan

Conversation

@Leechael

@Leechael Leechael commented Jun 2, 2026

Copy link
Copy Markdown
Collaborator

Adds a comprehensive operator guide for upgrading dstack-kms clusters across release boundaries, including a validated path to KMS 0.6.0.

Covers:

  • Cluster requirements and the onboard verification constraint
  • 0.6.0 upgrade matrix (validated): 0.5.3/0.5.4 → 0.5.7 → 0.6.0; 0.5.6+ → 0.6.0 direct
  • Critical: target KMS OS must not enable TDX lite attestation during upgrade — old sources only support legacy mode
  • RA-TLS OID break explaining why 0.5.3/0.5.4 cannot onboard 0.6.0 peers directly (No attestation provided)
  • Gateway upgrade order (after KMS) and VMM settings for the upgrade matrix
  • Pre-cutover checklist (allowlists, image download, key/CA continuity, rollback window)
  • Known historical release boundaries (0.5.4 → 0.5.5 → 0.5.9 → kms-v0.5.11) and bridge staircase
  • Pre-flight verification using dstack-mr diagnose
  • Risks, edge cases, and verifier capability matrix by release

Doc: docs/operations/kms-upgrade-plan.md

Leechael and others added 2 commits July 12, 2026 21:57
Comprehensive guide for upgrading dstack-kms clusters across release
boundaries (0.5.4 through kms-v0.5.11 + PR #693), covering:

- Cluster requirements and the onboard verification constraint
- Known release boundaries and recommended upgrade staircase
- Bridge upgrade approach for cross-boundary hops
- Pre-flight verification with dstack-mr diagnose
- Risks, edge cases, and verifier capability matrix
Incorporate validated operator conclusions for upgrading clusters to
KMS 0.6.0 (path matrix, RA-TLS OID break on 0.5.3/0.5.4, Gateway/VMM
order, pre-cutover checklist), and warn that the target KMS OS must
stay on legacy TDX attestation during onboard because older sources
only support legacy verification.
@kvinwang kvinwang force-pushed the docs/kms-upgrade-plan branch from 1c6f4eb to 203e09b Compare July 13, 2026 04:57
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.

2 participants