fix: validate machine config before activation

[alongubkin]

Background

Machine join renders configuration from a release manifest. TOML itself accepts values such as both true and "true", but Horizond requires specific scalar types. Replacing the active file before the actual consumer validates it can turn a repair or reconfiguration into a service crash loop.

Changes

• preserve manifest literal types when rendering TOML
• write configuration through a durable same-directory candidate file
• optionally run the bundle-declared configuration validator against that candidate
• atomically replace the active config only after validation succeeds
• leave the previous active config untouched and remove the candidate on rejection
• recover safely from stale candidate files left by an interrupted join

The validator is optional for backward compatibility with existing bundle manifests. Horizon PR #97 adds horizond validate-config <path> and publishes it in new manifests.

Validation

• cargo test -p alien-deploy-cli machine_config_ (4 passed)
• rejection test proves the old active file remains byte-for-byte unchanged

[greptile-apps]

Greptile Summary

The PR validates rendered machine configuration before activation and preserves literal TOML scalar types.

• Writes and durably syncs a same-directory candidate before atomically replacing the active configuration.
• Runs an optional bundle-provided validator and removes rejected candidates without altering the active file.
• Removes a stale candidate associated with a reused process ID before creating the new candidate.

Confidence Score: 5/5

The PR appears safe to merge.

No blocking failure remains; the previously reported stale-candidate issue is fixed for the regular files that interrupted candidate writes can leave behind.

Important Files Changed

Filename	Overview
crates/alien-deploy-cli/src/commands/join.rs	Adds typed TOML rendering, optional pre-activation validation, durable candidate writes, atomic replacement, stale-candidate recovery, and focused regression tests.

Flowchart

%%{init: {'theme': 'neutral'}}%%
flowchart LR
  A[Render typed TOML] --> B[Remove stale PID candidate]
  B --> C[Create and sync candidate]
  C --> D{Validator configured?}
  D -- No --> F[Atomically replace active config]
  D -- Yes --> E{Validation succeeds?}
  E -- Yes --> F
  E -- No --> G[Remove candidate]
  G --> H[Keep active config unchanged]
  F --> I[Sync parent directory]

Loading

_{Reviews (4): Last reviewed commit: "fix: validate machine config before acti..." | Re-trigger Greptile}