Skip to content

Harden universe A2A reachability probe#4621

Merged
asim merged 1 commit into
masterfrom
codex/increment-4620
Jul 10, 2026
Merged

Harden universe A2A reachability probe#4621
asim merged 1 commit into
masterfrom
codex/increment-4620

Conversation

@asim

@asim asim commented Jul 10, 2026

Copy link
Copy Markdown
Member

Summary

  • Retry the universe harness A2A reachability probe within the caller deadline so transient gateway/provider latency does not turn a reachable concierge agent into a false negative.
  • Keep the probe side-effect safe by reusing the explicit no-notification prompt and preserving the post-probe notification-count assertion.
  • Add a regression test that forces a first A2A request timeout and verifies the probe retries successfully.

Closes #4504
Closes #4620

Testing

  • go test ./internal/harness/universe
  • go build ./...
  • go test ./... (fails in this sandbox: AtlasCloud stream conformance received live 400, and gRPC loopback targets are forbidden by envoy)
  • golangci-lint run ./... (fails in this sandbox: installed golangci-lint was built with Go 1.24 while repo targets Go 1.25.12)

@asim asim added the codex label Jul 10, 2026
Copilot AI review requested due to automatic review settings July 10, 2026 21:59
@asim
asim merged commit 29d8544 into master Jul 10, 2026
13 checks passed

Copilot AI left a comment

Copy link
Copy Markdown

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Pull request overview

This PR hardens the universe harness A2A reachability probe to reduce false negatives caused by transient A2A gateway/provider latency, while preserving the probe’s side-effect safety guarantees (no duplicate notifications). It also adds a regression test to validate retry behavior.

Changes:

  • Add retry/backoff logic to the A2A reachability probe, keeping retries within the caller’s deadline.
  • Preserve the side-effect-safe probe prompt and keep the “no extra notifications” assertion around the probe.
  • Add a regression test that simulates a transient failure on the first A2A request and verifies a successful retry.

Reviewed changes

Copilot reviewed 2 out of 2 changed files in this pull request and generated 3 comments.

File Description
internal/harness/universe/main.go Adds deadline-aware retry/backoff behavior to the A2A reachability probe.
internal/harness/universe/main_test.go Adds a regression test covering “first attempt fails, second attempt succeeds” behavior for the probe.

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

if time.Until(deadline) <= 0 {
return fmt.Errorf("A2A reachability probe failed after %d attempt(s): %w", attempt, lastErr)
}
time.Sleep(minDuration(200*time.Millisecond*time.Duration(attempt), time.Until(deadline)))
Comment on lines +34 to +36
if got, want := r.URL.Path, "/agents/concierge"; got != want {
t.Fatalf("path = %q, want %q", got, want)
}
Comment on lines +37 to +40
if atomic.AddInt64(&calls, 1) == 1 {
time.Sleep(5 * time.Second)
return
}
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

Projects

None yet

Development

Successfully merging this pull request may close these issues.

Loop: build increment #186 Make AtlasCloud universe A2A reachability probe deterministic

3 participants