Repository navigation
persist sessionId on failure so a failed run stays resumable - #47
Closed
matt-wright86 wants to merge 517 commits into
Closed
matt-wright86 wants to merge 517 commits into
matt-wright86 wants to merge 517 commits into
Conversation
- Add CreateTable method to ProjectionStore interface - Add RegisteredTables method to ProjectionEngine interface - Implement auto-create table in Register() method - Track registered tables for rebuild command - Add tests for table creation and RegisteredTables() Closes bf-ca98.4
- Add AccountDirectoryProjector that maintains pats array instead of pat_count - PATCount derived from len(pats) - no auxiliary keys needed - Idempotent handling for all events including duplicate PATs - Add CreateTable method to mockProjectionStore in test helpers
Implements projector for PATCreated (insert) and PATRevoked (delete) events.
Table: projection_pat_by_id
Key: pat_id
Doc: {pat_id, key_hash, account_id}
Implements RealmDirectoryProjector for the projection refactor. - Handles RealmCreated events (insert with status active) - Handles RealmSuspended events (read-modify-write to suspended) - Table: projection_realm_directory - Key: realm_id
…solution
- Add RealmNameLookupProjector with projection_realm_name_lookup table
- Key: name, Doc: {name, realm_id}
- Handles RealmCreated event with idempotency
- Enables O(1) realm-name-to-ID resolution instead of scanning
Implements username to account_id lookup projector:
- Table: projection_username_lookup
- Key: username
- Document: {username, account_id}
- Handles AccountCreated event with idempotency
Rune: bf-ca98.14
Write DependencyExistenceProjector to maintain a table where each row
represents a single dependency. Row existence is the answer to whether
a dependency exists.
- Table: projection_dependency_existence
- Key: {rune_id}:{target_id}:{relationship}
- Events: DependencyAdded (insert), DependencyRemoved (delete)
- Skips inverse events (handled by forward events)
This replaces the dep:* boolean keys from the old dependency_graph projector.
Implements projector for tracking system-wide status: - Tracks admin/owner accounts in _admin realm (admin_account_ids) - Tracks existing realms (realm_ids) - Single-row-per-realm invariant with key 'status' - Handles AccountCreated, RoleAssigned, RoleRevoked, RealmCreated - All handlers are idempotent
Implement projector that maintains cycle-check edges in projection_dependency_cycle_check table.
- Handles DependencyAdded (insert) and DependencyRemoved (delete) events
- Key: {source_id}:{target_id}, Doc: {source_id, target_id}
- Row existence indicates cycle would occur if edge is added
- Skips inverse events, handles unknown events gracefully
Define ScriptContext, ScriptTaskDefinition, ScriptResult, and ExecutionStats in interfaces-task as the foundation contract for runner and agent packages.
… (GH #33). Introduces ed25519 signing with deterministic JSON canonicalization, direction-agnostic RPC frames, and peer APIs where the runner dials and both sides communicate via subscribe/send. The orchestrator multiplexes connected runners through onPeerConnect/onPeerDisconnect and send(peerId), with loopback vitest-gwt specs covering crypto, bidirectional RPC, streaming, and tamper rejection.
Merge interfaces-task and interfaces-task-source into @bifrost-ai/interfaces-work. Replace the split Task/Script concepts with WorkItem instances and WorkItemHandlers. Handlers receive the work item and an execution context separately; RPC methods and runner registration APIs use the same vocabulary. Reserve "task" for the LLM task agent (kind: "task").
unify orchestrator v2 execution model under work item terminology
runTaskAgent persisted the engine sessionId only after the success/failure branch, so a failed run lost its session and a retry had to rebuild context from scratch. Persist it before the branch, and return telemetry on the failure path too. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Contributor
📝 WalkthroughSummary by CodeRabbit
WalkthroughThe ChangesSession persistence and telemetry on failure
Possibly related PRs
🚥 Pre-merge checks | ✅ 5✅ Passed checks (5 passed)
Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out. Comment |
![coderabbitai[bot]](https://avatars.githubusercontent.com/in/347564?s=60&v=4){kind=small}
Contributor
There was a problem hiding this comment.
Actionable comments posted: 1
🤖 Prompt for all review comments with AI agents
Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.
Inline comments:
In `@orchestrator-v2/packages/agent-3-task/src/run-task-agent.ts`:
- Around line 51-58: The unguarded ctx.setState call in runTaskAgent can now
fail the whole function on both success and failure paths, so wrap the
persistence step in try/catch and keep returning a structured WorkItemResult if
state saving rejects. Use the existing runTaskAgent flow and the
engineResult.sessionId / ctx.setState block to locate the change, and make sure
any persistence error is handled without throwing past the function boundary.
🪄 Autofix (Beta)
Fix all unresolved CodeRabbit comments on this PR:
- Push a commit to this branch (recommended)
- Create a new PR with the fixes
ℹ️ Review info
⚙️ Run configuration
Configuration used: Repository UI
Review profile: CHILL
Plan: Pro
Run ID: b73d428d-7978-42eb-89b0-4759a9fc82f0
📒 Files selected for processing (2)
orchestrator-v2/packages/agent-3-task/src/run-task-agent.spec.tsorchestrator-v2/packages/agent-3-task/src/run-task-agent.ts
Comment on lines
+51
to
58
| // Persist the sessionId before branching on the outcome, so a failed run stays | ||
| // resumable — a retry can --resume instead of rebuilding context from scratch. | ||
| if (engineResult.sessionId !== undefined) { | ||
| await ctx.setState({ | ||
| ...workItem.state, | ||
| sessionId: engineResult.sessionId, | ||
| }); | ||
| } |
Contributor
There was a problem hiding this comment.
🩺 Stability & Availability | 🟡 Minor | ⚡ Quick win
Unhandled ctx.setState failure now impacts both success and failure paths.
This persistence call isn't wrapped in try/catch. If ctx.setState rejects, the function throws instead of returning a structured WorkItemResult, defeating the purpose of making failed runs resumable. Since this block now runs unconditionally (previously only on the success path), the blast radius of an unhandled rejection here has doubled.
🛡️ Suggested guard
if (engineResult.sessionId !== undefined) {
- await ctx.setState({
- ...workItem.state,
- sessionId: engineResult.sessionId,
- });
+ try {
+ await ctx.setState({
+ ...workItem.state,
+ sessionId: engineResult.sessionId,
+ });
+ } catch {
+ // Swallow persistence errors so a failing setState doesn't crash the run;
+ // resumability is best-effort and shouldn't mask the underlying engine result.
+ }
}📝 Committable suggestion
‼️ IMPORTANT
Carefully review the code before committing. Ensure that it accurately replaces the highlighted code, contains no missing lines, and has no issues with indentation. Thoroughly test & benchmark the code to ensure it meets the requirements.
Suggested change
| // Persist the sessionId before branching on the outcome, so a failed run stays | |
| // resumable — a retry can --resume instead of rebuilding context from scratch. | |
| if (engineResult.sessionId !== undefined) { | |
| await ctx.setState({ | |
| ...workItem.state, | |
| sessionId: engineResult.sessionId, | |
| }); | |
| } | |
| // Persist the sessionId before branching on the outcome, so a failed run stays | |
| // resumable — a retry can --resume instead of rebuilding context from scratch. | |
| if (engineResult.sessionId !== undefined) { | |
| try { | |
| await ctx.setState({ | |
| ...workItem.state, | |
| sessionId: engineResult.sessionId, | |
| }); | |
| } catch { | |
| // Swallow persistence errors so a failing setState doesn't crash the run; | |
| // resumability is best-effort and shouldn't mask the underlying engine result. | |
| } | |
| } |
🤖 Prompt for AI Agents
Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.
In `@orchestrator-v2/packages/agent-3-task/src/run-task-agent.ts` around lines 51
- 58, The unguarded ctx.setState call in runTaskAgent can now fail the whole
function on both success and failure paths, so wrap the persistence step in
try/catch and keep returning a structured WorkItemResult if state saving
rejects. Use the existing runTaskAgent flow and the engineResult.sessionId /
ctx.setState block to locate the change, and make sure any persistence error is
handled without throwing past the function boundary.
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
runTaskAgentpersisted the enginesessionIdonly after the success/failurebranch, so a failed run lost its session — a retry had to rebuild context from
scratch. Persist it before the branch (failed runs stay resumable), and
return telemetry on the failure path too.
Type of change
Checklist
go/npxwere not usedmakegates — N/A:orchestrator-v2uses its ownvptoolchainvp run -r buildclean ·vp run -r testgreensessionIdpersisted + telemetry returned)main/simplify)Notes for reviewers
First of the orchestrator-v2 hardening pieces (from the earlier experiment branch)
re-expressed onto the work-item model as focused PRs. Next up: I3 (callback
resilience), Route A (telemetry as a terminal outcome), I1 (capability routing).
🤖 Generated with Claude Code