Fix queue ActiveCount race in HandleActionUpdates terminal pathsb/compute-batch

NotifyQueueActionComplete (which decrements ActiveCount) was called after
releasing m_ResultsLock. GetActionResult acquires m_ResultsLock to consume
the result, so a caller could observe ActiveCount still at 1 immediately
after GetActionResult returned OK if the scheduler thread was preempted
between releasing m_ResultsLock and reaching NotifyQueueActionComplete.

Fix by calling NotifyQueueActionComplete before the m_ResultsLock block
that publishes the result into m_ResultsMap. This guarantees that by the
time GetActionResult can return OK, the queue counters are already updated.

1 files changed, 7 insertions, 1 deletions

diff --git a/src/zencompute/computeservice.cpp b/src/zencompute/computeservice.cpp
index 2e49c1114..43031955a 100644
--- a/src/zencompute/computeservice.cpp
+++ b/src/zencompute/computeservice.cpp
@@ -1896,6 +1896,13 @@ ComputeServiceSession::Impl::HandleActionUpdates()
 
 						RemoveActionFromActiveMaps(ActionLsn);
 
+						// Update queue counters BEFORE publishing the result into
+						// m_ResultsMap. GetActionResult erases from m_ResultsMap
+						// under m_ResultsLock, so if we updated counters after
+						// releasing that lock, a caller could observe ActiveCount
+						// still at 1 immediately after GetActionResult returned OK.
+						NotifyQueueActionComplete(Action->QueueId, ActionLsn, TerminalState);
+
 						m_ResultsLock.WithExclusiveLock([&] {
 							m_ResultsMap[ActionLsn] = Action;
 
@@ -1927,7 +1934,6 @@ ComputeServiceSession::Impl::HandleActionUpdates()
 								  Action->ActionId,
 								  ActionLsn,
 								  TerminalState == RunnerAction::State::Completed ? "SUCCESS" : "FAILURE");
-						NotifyQueueActionComplete(Action->QueueId, ActionLsn, TerminalState);
 						break;
 					}
 			}