From 7444b52763908b9a042699e792fbad4e630b3955 Mon Sep 17 00:00:00 2001
From: operator <operator@netvm.local>
Date: Tue, 6 Oct 2026 03:14:02 +0000
Subject: [PATCH] Fix split-brain swarm dispatch: narrow harvester
 SWARM_WORKER_POOL to [muse]

The 2026-10-05 fix note claimed dev/def were removed from the pool but the
code still listed [dev, def, muse]. The harvester's every-minute systemd
dispatch raced the pool daemon claiming slots as dev/def, which refuse on
attribution grounds (dev FAILs fast, def freezes until the 60-min reaper) -
the fleet-wide dev-FAIL-fast + def-frozen pattern on every swarm.

Pool now matches swarm_worker/daemon.py WORKER_POOL exactly: muse only,
the single fully authenticated auxiliary worker. Reporting/harvest paths
untouched.
---
 bin/response-harvester.py | 18 ++++++++++++++++--
 1 file changed, 16 insertions(+), 2 deletions(-)

diff --git a/bin/response-harvester.py b/bin/response-harvester.py
index 3cb25e1..17109d4 100755
--- a/bin/response-harvester.py
+++ b/bin/response-harvester.py
@@ -960,6 +960,13 @@ def archive_ephemeral_thread(agent, thread_id, job_id=None):
     except Exception as e:
         sys.stderr.write(f"warning: failed to update job-sidechats state: {e}\n")
 
+    # Complete subagent tracker record if this thread corresponds to an ephemeral subagent session
+    try:
+        import subagent_tracker
+        subagent_tracker.complete_session(thread_id, note=f"Harvested verdict for job {job_id}")
+    except Exception:
+        pass
+
     # Call muse-threads.py archive via subprocess (runs in node's netns)
     try:
         helper = BIN_DIR / "muse-threads.py"
@@ -975,7 +982,13 @@ def archive_ephemeral_thread(agent, thread_id, job_id=None):
 # NOTE 2026-10-05 (Fix Agent 2/5): dev/def removed from the pool. No worker
 # agents exist on dev/def (no Meta sessions provisioned), so slots dispatched
 # to them froze with null results. Re-add only after real dev/def workers exist.
-SWARM_WORKER_POOL = ["dev", "def", "muse"]
+# NOTE 2026-10-06 (split-brain dispatch fix): the removal above was comment-only;
+# the code still listed dev/def, and the harvester's every-minute dispatch raced
+# the pool daemon claiming slots as dev/def, which refuse on attribution grounds
+# (dev FAILs fast, def freezes until the 60-min reaper). Pool now matches
+# swarm_worker/daemon.py WORKER_POOL exactly: only muse, the single fully
+# authenticated auxiliary worker.
+SWARM_WORKER_POOL = ["muse"]
 
 # Stuck-slot reaper: a slot that stays "running" with no result longer than
 # this is treated as wedged (worker died / dispatch lost). Healthy slots
@@ -1003,7 +1016,8 @@ def reconcile_and_dispatch_swarms(dry_run=False):
     """
     Autonomous Swarm Orchestrator:
     1. Scans swarms.json for pending slots.
-    2. Dynamically allocates available auxiliary worker nodes (dev, def, muse).
+    2. Dynamically allocates the auxiliary worker pool (muse only; dev/def have
+    no provisioned worker agents and refuse on attribution grounds).
     3. Provisions ephemeral sidechat per slot and dispatches the task with [RESULT <swarm_id>/<slot>].
     4. Upon completion of all slots, sends completion summary DM to originating coordinator.
     """