Skip to content

docs: update BRC timeout wait for event-driven polling [doc-updater] #1934

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
jwbron merged 2 commits into from
Apr 23, 2026
Merged

docs: update BRC timeout wait for event-driven polling [doc-updater] #1934

Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
745bdea
docs: update BRC timeout wait description for event-driven polling
jwbron Apr 23, 2026
7c3184b
docs: split long timeout-wait paragraph into bulleted list for readab…
egg-reviewer[bot] Apr 23, 2026
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
8 changes: 7 additions & 1 deletion docs/guides/concurrent-execution.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -701,7 +701,13 @@ If the BRC tracker is unavailable, the orchestrator falls back to the old behavi

Timeout handling is idempotent — if the timeout fires multiple times (e.g., due to a race with the overseer), only the first invocation takes effect.

**Consensus reached during timeout wait**: After the BRC timeout evaluation, the orchestrator waits for still-running containers to exit. If consensus becomes complete during this window (all agents confirmed, no unresolved NACKs), the orchestrator recovers: it marks agents complete, restores the pipeline from `FAILED` to `RUNNING` if needed, and returns success — the timeout evaluation is overridden by the consensus outcome. If consensus is complete but unresolved NACKs remain, the orchestrator escalates to HITL with options "Retry phase", "Accept current state", "Abort phase". See [issue #1693](https://github.com/jwbron/egg/issues/1693).
**Consensus reached during timeout wait**: After the BRC timeout evaluation, the orchestrator enters an event-driven polling loop that rechecks consensus proactively:

- **Polling mechanics**: 30-second intervals, up to 3600s total budget.
- **Consensus complete** (all agents confirmed, no unresolved NACKs): the orchestrator immediately stops remaining containers, marks agents complete, restores the pipeline from `FAILED` to `RUNNING` if needed, and returns success — the timeout evaluation is overridden by the consensus outcome.
- **Unresolved NACKs remain**: the orchestrator escalates to HITL with options "Retry phase", "Accept current state", "Abort phase". See [issue #1693](https://github.com/jwbron/egg/issues/1693).

This replaces the former per-container blocking wait, closing the race where a late NACK→re-propose→ACK cycle completing near the end of the budget could previously be missed ([issue #1921](https://github.com/jwbron/egg/issues/1921)).

### Consensus Stall Recovery

Expand Down
Loading