Skip to content

sqlite: persist device parents and execution allocations - #962

Draft
zibi-pplx wants to merge 1 commit into
pr960from
pr943
Draft

zibi-pplx wants to merge 1 commit into
pr960from
pr943

Conversation

@zibi-pplx

@zibi-pplx zibi-pplx commented Oct 26, 2026 •

Copy link
Copy Markdown
Collaborator

Summary

Add node-local device enrollment, membership, allocation and active-claim tables, plus persisted sandbox device requirements. Separate physical parents from execution allocations so whole-device ownership does not require one owner field on the parent. Composite foreign keys enforce membership, and exclusive claims prevent two active exclusive holders of the same member.

Validation

Device-store tests exercised fenced admission, release, reset recovery, ownership snapshots and rollback guards during implementation. The migrations were applied on the dedicated development node used for the GPU lifecycle smoke test. Full workspace testing remains with CI.

Severity and risk

New additive SQLite schema. Deploy before the consumers in this stack. Down migrations refuse while enrollment/ownership history or sandbox requirements remain; binary rollback cannot silently turn device-bearing work into CPU-only work. No control-plane PostgreSQL change or automatic device enrollment.

Stack

GPU execution, after the CPU-only Cloud Hypervisor group. #962 is the first PR in this group and supplies the SQLite schema before #945 and the device consumers. Requirements, artifact delivery, fenced ownership, operator visibility and GPU smoke coverage follow. Device-bearing requests remain standalone-only. The scrub candidate in #950 is unqualified; this stack does not claim production-safe tenant reuse.

Earlier review discussion remains in #944; this PR carries the same implementation in its new stack position.

AI assistance: Codex implemented and validated the changes and drafted this description.


Stack created with Sapling. Best reviewed with ReviewStack.
Base: #924, the separate Cloud Hypervisor installation stack.

@zibi-pplx
zibi-pplx added this pull request to stack #963 October 26, 2026 03:04
This was referenced Oct 26, 2026
## Summary

Add node-local device enrollment, membership, allocation and active-claim tables, plus persisted sandbox device requirements. Separate physical parents from execution allocations so whole-device ownership does not require one owner field on the parent. Composite foreign keys enforce membership, and exclusive claims prevent two active exclusive holders of the same member.

## Validation

Device-store tests exercised fenced admission, release, reset recovery, ownership snapshots and rollback guards during implementation. The migrations were applied on the dedicated development node used for the GPU lifecycle smoke test. Full workspace testing remains with CI.

## Severity and risk

New additive SQLite schema. Deploy before the consumers in this stack. Down migrations refuse while enrollment/ownership history or sandbox requirements remain; binary rollback cannot silently turn device-bearing work into CPU-only work. No control-plane PostgreSQL change or automatic device enrollment.

## Stack

GPU execution, after the CPU-only Cloud Hypervisor group. #962 is the first PR in this group and supplies the SQLite schema before #945 and the device consumers. Requirements, artifact delivery, fenced ownership, operator visibility and GPU smoke coverage follow. Device-bearing requests remain standalone-only. The scrub candidate in #950 is unqualified; this stack does not claim production-safe tenant reuse.

Earlier review discussion remains in #944; this PR carries the same implementation in its new stack position.

AI assistance: Codex implemented and validated the changes and drafted this description.

This branch has not been deployed

No deployments
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.

1 participant