From 2bd97791c963b09fd01eee622c55a344e46cfeb5 Mon Sep 17 00:00:00 2001
From: saravanakumardb1 <saravanakumardb1@users.noreply.github.com>
Date: Sun, 31 May 2026 16:13:52 -0700
Subject: [PATCH] feat(fleet): resilient PR merge on ship (inline attempt +
 background retry)

The corporate proxy intermittently 407s GitHub's API, so a single gh pr merge can
fail transiently. Try once inline (fast path), then retry in the background with
backoff (3s/8s/20s/45s) without blocking the ship; mark prState=merged when one
lands. Best-effort throughout.
---
 .../src/modules/fleet/coordinator.ts          | 41 +++++++++++++++----
 1 file changed, 34 insertions(+), 7 deletions(-)

diff --git a/services/platform-service/src/modules/fleet/coordinator.ts b/services/platform-service/src/modules/fleet/coordinator.ts
index 3d985711..8f46566f 100644
--- a/services/platform-service/src/modules/fleet/coordinator.ts
+++ b/services/platform-service/src/modules/fleet/coordinator.ts
@@ -643,17 +643,44 @@ const execFileAsync = promisify(execFile);
  * authenticated (the coordinator host). Best-effort: a failure leaves the PR open
  * and never fails the ship. Marks the run prState=merged on success.
  */
+async function ghMergePr(prUrl: string): Promise<boolean> {
+  try {
+    await execFileAsync('gh', ['pr', 'merge', prUrl, '--squash', '--delete-branch'], {
+      timeout: 60_000,
+    });
+    return true;
+  } catch {
+    // Network/proxy hiccup (e.g. intermittent 407), checks pending, or conflict.
+    return false;
+  }
+}
+
 async function mergeRunPrOnShip(jobId: string, latest: FleetRunDoc | undefined): Promise<void> {
   if (process.env.FLEET_SHIP_MERGES_PR !== '1') return;
   if (!latest?.prUrl || latest.prState === 'merged') return;
-  try {
-    await execFileAsync('gh', ['pr', 'merge', latest.prUrl, '--squash', '--delete-branch'], {
-      timeout: 60_000,
-    });
-    await repo.updateRun(latest.id, jobId, { prState: 'merged' });
-  } catch {
-    // best-effort — leave the PR open if the merge is blocked (checks, conflicts…)
+  const { prUrl, id: runId } = latest;
+  // Fast attempt inline (so a healthy proxy merges immediately without delaying ship).
+  if (await ghMergePr(prUrl)) {
+    await repo.updateRun(runId, jobId, { prState: 'merged' });
+    return;
   }
+  // The corporate proxy intermittently 407s GitHub's API. Retry in the BACKGROUND
+  // with backoff so the ship response is never blocked; mark merged when one lands.
+  void (async () => {
+    for (const delayMs of [3_000, 8_000, 20_000, 45_000]) {
+      await new Promise(r => setTimeout(r, delayMs));
+      if (await ghMergePr(prUrl)) {
+        try {
+          await repo.updateRun(runId, jobId, { prState: 'merged' });
+        } catch {
+          /* run gone — ignore */
+        }
+        return;
+      }
+    }
+  })().catch(() => {
+    /* detached best-effort — never throws */
+  });
 }
 
 export async function patchJobFenced(