SERVER-17364 Don't stash RecoveryUnits across getMores

We now tell PlanExecutors to detach from their OperationContexts and to shed
all storage engine resources before stashing the ClientCursor. This is a
heavier weight operation than a normal save/restoreState which is no longer
allowed to change the OperationContext.

1 files changed, 5 insertions, 1 deletions

diff --git a/src/mongo/db/commands/repair_cursor.cpp b/src/mongo/db/commands/repair_cursor.cpp
index eef18cfcfeb..5f2f88c04ca 100644
--- a/src/mongo/db/commands/repair_cursor.cpp
+++ b/src/mongo/db/commands/repair_cursor.cpp
@@ -103,11 +103,15 @@ public:
         // it now so that it can be registed with ClientCursor.
         exec->deregisterExec();
         exec->saveState();
+        exec->detachFromOperationContext();
 
         // ClientCursors' constructor inserts them into a global map that manages their
         // lifetimes. That is why the next line isn't leaky.
         ClientCursor* cc =
-            new ClientCursor(collection->getCursorManager(), exec.release(), ns.ns());
+            new ClientCursor(collection->getCursorManager(),
+                             exec.release(),
+                             ns.ns(),
+                             txn->recoveryUnit()->isReadingFromMajorityCommittedSnapshot());
 
         appendCursorResponseObject(cc->cursorid(), ns.ns(), BSONArray(), &result);