feat(pb): actor v1 to v2 migration

[MasterPtato]

Description

Please include a summary of the changes and the related issue. Please also include relevant motivation and context.

Type of change

• Bug fix (non-breaking change which fixes an issue)
• New feature (non-breaking change which adds functionality)
• Breaking change (fix or feature that would cause existing functionality to not work as expected)
• This change requires a documentation update

How Has This Been Tested?

Please describe the tests that you ran to verify your changes.

Checklist:

• My code follows the style guidelines of this project
• I have performed a self-review of my code
• I have commented my code, particularly in hard-to-understand areas
• I have made corresponding changes to the documentation
• My changes generate no new warnings
• I have added tests that prove my fix is effective or that my feature works
• New and existing unit tests pass locally with my changes

[railway-app]

🚅 Deployed to the rivet-pr-4548 environment in rivet-frontend

Service	Status	Web	Updated (UTC)
frontend-cloud	❌ Build Failed (View Logs)	Web	Apr 3, 2026 at 3:42 am
website	❌ Build Failed (View Logs)	Web	Apr 3, 2026 at 3:42 am
mcp-hub	✅ Success (View Logs)	Web	Apr 3, 2026 at 3:42 am
frontend-inspector	❌ Build Failed (View Logs)	Web	Apr 3, 2026 at 3:42 am
ladle	❌ Build Failed (View Logs)	Web	Apr 3, 2026 at 3:41 am
kitchen-sink	❌ Build Failed (View Logs)	Web	Apr 3, 2026 at 3:41 am

[MasterPtato]

Warning

This pull request is not mergeable via GitHub because a downstack PR is open. Once all requirements are satisfied, merge this PR as a stack on Graphite.
Learn more

• feat: native SQLite backend via napi-rs with KV channel protocol #4496
• fix: optimize e2e actor path #4551
• feat(pb): actor v1 to v2 migration #4548 👈 (View in Graphite)
• feat(envoy-client): fully flesh out tunnel impl #4547
• feat: basic rivetkit impl #4544
• feat: basic envoy tunnel impl #4534
• chore: envoy client #4521 : 1 other dependent PR (#4531 )
• feat(pb): actors v3 #4463
• feat(ups): implement queue subs #4486
• feat(cache): add in flight deduping #4459
• fix(cache): clean up lib #4457
• fix(config): allow configuring topo dcs via map, fix pg ssl mode config #4456
• fix(api): dont 500 on cross dc response #4449
• chore: update byte units on frontend #4444
• fix: runner alloc idx logic, api auth for actor get #4443
• main

This stack of pull requests is managed by Graphite. Learn more about stacking.

[claude]

PR Review: feat(pb): actor v1 to v2 migration

This PR implements actor v1 to v2 migration with a clear separation of the Reallocating state from Sleeping, along with several correctness fixes. Overall the approach is solid. A few things to flag:

Bugs / Correctness

1. Typo in description (double "before") — config-schema.json and config/pegboard.rs

The description reads: "How long to wait after starting to attempt to reallocate before before setting actor to sleep." The word before appears twice.

2. Reallocating to Sleeping transition logic — actor2/mod.rs

The check state.retry_backoff_state.last_retry_ts > *since_ts + actor_retry_duration_threshold() checks whether the last retry occurred more than threshold ms after since_ts. But if last_retry_ts is stale from a previous Reallocating epoch, the comparison could behave unexpectedly. If last_retry_ts is 0/unset, this never triggers. The intent seems to be "have we been retrying for more than N ms since entering this state", which might be better expressed as now > *since_ts + threshold with now captured via a GetTs activity for determinism. Worth validating against actual last_retry_ts update semantics.

3. Potential race in gateway v1 migration path — pegboard_gateway.rs

When migrate_sub.next() fires and we delegate to handle_actor_v2, the v1 workflow has dispatched v2 and emitted MigratedToV2, but may still be in the process of returning. Worth confirming that v1's pegboard_actor returns Ok(()) immediately after dispatching v2 without any further envoy interaction, so there is no brief window of contention between both workflows.

Style / Minor

4. Inconsistent now sourcing in reschedule_actor — runtime.rs

allocate now returns AllocateOutput { allocation, now }, and handle_stopped correctly uses allocate_res.now. But reschedule_actor then calls a fresh GetTs activity after calling allocate internally. Since allocate returns now in its output, it could be reused to avoid an extra activity replay step.

Positive changes worth noting

• Determinism fix: Replacing all direct util::timestamp::now() calls inside workflow code with ctx.activity(runtime::GetTsInput {}) is the right approach for workflow replay determinism. Good catch throughout mod.rs and runtime.rs.
• Reallocating state extraction: Splitting Sleeping { attempting_reallocation: bool } into distinct Sleeping and Reallocating { since_ts } variants is a meaningful clarity improvement.
• handleShutdown idempotency guard: The shuttingDown flag cleanly prevents double-shutdown.
• res bug fix in connection.ts: The !res && !errored guard correctly prevents overwriting a previously-set result when a WebSocket error breaks the loop.
• Missing .tag("actor_id", ...): Adding the actor ID tag when dispatching the v2 workflow from v1 is critical for correct workflow routing.
• API endpoint fix: /actors2 to /actors in the remote manager driver was clearly a bug.
• Runner pool metadata namespace fix: keys::subspace() to namespace::keys::subspace() in runner_pool_metadata_poller.rs looks like an important correctness fix.

Checklist (draft PR)

• PR description template is unfilled — tests, motivation, and checklist should be completed before marking ready for review.
• scripts/release/update_version.ts should be verified since the OpenAPI version bumped from 2.2.0 to 2.2.1.