defer creating the id index on a clone until we first insert all the data

4 files changed, 42 insertions, 12 deletions

diff --git a/db/cloner.cpp b/db/cloner.cpp
index c23a3d5f32b..a84b238fee5 100644
--- a/db/cloner.cpp
+++ b/db/cloner.cpp
@@ -165,6 +165,9 @@ namespace mongo {
         }
     }
     
+    extern bool inDBRepair;
+    void ensureIdIndexForNewNs(const char *ns);
+
     bool Cloner::go(const char *masterHost, string& errmsg, const string& fromdb, bool logForRepl, bool slaveOk, bool useReplAuth, bool snapshot) {
 
 		massert( 10289 ,  "useReplAuth is not written to replication log", !useReplAuth || !logForRepl );
@@ -256,19 +259,37 @@ namespace mongo {
             assert(p);
             string to_name = todb + p;
 
+            bool wantIdIndex = false;
             {
                 string err;
                 const char *toname = to_name.c_str();
-                userCreateNS(toname, options, err, logForRepl);
+                /* we defer building id index for performance - building it in batch is much faster */ 
+                userCreateNS(toname, options, err, logForRepl, &wantIdIndex);
             }
             log(1) << "\t\t cloning " << from_name << " -> " << to_name << endl;
             Query q;
             if( snapshot ) 
                 q.snapshot();
             copy(from_name, to_name.c_str(), false, logForRepl, masterSameProcess, slaveOk, q);
+
+            if( wantIdIndex ) {
+                /* we need dropDups to be true as we didn't do a true snapshot and this is before applying oplog operations 
+                   that occur during the initial sync.  inDBRepair makes dropDups be true.
+                   */
+                bool old = inDBRepair;
+                try {
+                    inDBRepair = true;
+                    ensureIdIndexForNewNs(to_name.c_str());
+                }
+                catch(...) { 
+                    inDBRepair = old;
+                    throw;
+                }
+            }
         }
 
         // now build the indexes
+
         string system_indexes_from = fromdb + ".system.indexes";
         string system_indexes_to = todb + ".system.indexes";
         /* [dm]: is the ID index sometimes not called "_id_"?  There is other code in the system that looks for a "_id" prefix 
diff --git a/db/oplog.cpp b/db/oplog.cpp
index 71cb8863c44..c2727d64970 100644
--- a/db/oplog.cpp
+++ b/db/oplog.cpp
@@ -231,7 +231,7 @@ namespace mongo {
     }
 
     /*@ @param opstr:
-          c userCreateNs
+          c userCreateNS
           i insert
           n no-op / keepalive
           d delete / remove
diff --git a/db/pdfile.cpp b/db/pdfile.cpp
index d00c51545cb..2da558c5f82 100644
--- a/db/pdfile.cpp
+++ b/db/pdfile.cpp
@@ -123,7 +123,7 @@ namespace mongo {
     void ensureIdIndexForNewNs(const char *ns) {
         if ( ( strstr( ns, ".system." ) == 0 || legalClientSystemNS( ns , false ) ) &&
              strstr( ns, ".$freelist" ) == 0 ){
-            log( 1 ) << "adding _id index for new collection" << endl;
+            log( 1 ) << "adding _id index for collection " << ns << endl;
             ensureHaveIdIndex( ns );
         }        
     }
@@ -161,7 +161,7 @@ namespace mongo {
         return z;
     }
 
-    bool _userCreateNS(const char *ns, const BSONObj& options, string& err) {
+    bool _userCreateNS(const char *ns, const BSONObj& options, string& err, bool *deferIdIndex) {
         if ( nsdetails(ns) ) {
             err = "collection already exists";
             return false;
@@ -223,15 +223,22 @@ namespace mongo {
         NamespaceDetails *d = nsdetails(ns);
         assert(d);
 
+        bool ensure = false;
         if ( options.getField( "autoIndexId" ).type() ) {
             if ( options["autoIndexId"].trueValue() ){
-                ensureIdIndexForNewNs( ns );
+                ensure = true;
             }
         } else {
             if ( !newCapped ) {
-                ensureIdIndexForNewNs( ns );
+                ensure=true;
             }
         }
+        if( ensure ) { 
+            if( deferIdIndex )
+                *deferIdIndex = true;
+            else
+                ensureIdIndexForNewNs( ns );
+        }
 
         if ( mx > 0 )
             d->max = mx;
@@ -239,14 +246,16 @@ namespace mongo {
         return true;
     }
 
-    // { ..., capped: true, size: ..., max: ... }
-    // returns true if successful
-    bool userCreateNS(const char *ns, BSONObj options, string& err, bool logForReplication) {
+    /** { ..., capped: true, size: ..., max: ... }
+        @param deferIdIndex - if not not, defers id index creation.  sets the bool value to true if we wanted to create the id index.
+        @return true if successful
+    */
+    bool userCreateNS(const char *ns, BSONObj options, string& err, bool logForReplication, bool *deferIdIndex) {
         const char *coll = strchr( ns, '.' ) + 1;
         massert( 10356 ,  "invalid ns", coll && *coll );
         char cl[ 256 ];
         nsToDatabase( ns, cl );
-        bool ok = _userCreateNS(ns, options, err);
+        bool ok = _userCreateNS(ns, options, err, deferIdIndex);
         if ( logForReplication && ok ) {
             if ( options.getField( "create" ).eoo() ) {
                 BSONObjBuilder b;
@@ -679,7 +688,7 @@ namespace mongo {
             NamespaceDetails *freeExtents = nsdetails(s.c_str());
             if( freeExtents == 0 ) { 
                 string err;
-                _userCreateNS(s.c_str(), BSONObj(), err);
+                _userCreateNS(s.c_str(), BSONObj(), err, 0);
                 freeExtents = nsdetails(s.c_str());
                 massert( 10361 , "can't create .$freelist", freeExtents);
             }
diff --git a/db/pdfile.h b/db/pdfile.h
index 5cef15bd927..0c1293e6f19 100644
--- a/db/pdfile.h
+++ b/db/pdfile.h
@@ -48,7 +48,7 @@ namespace mongo {
     
     /* deletes this ns, indexes and cursors */
     void dropCollection( const string &name, string &errmsg, BSONObjBuilder &result ); 
-    bool userCreateNS(const char *ns, BSONObj j, string& err, bool logForReplication);
+    bool userCreateNS(const char *ns, BSONObj j, string& err, bool logForReplication, bool *deferIdIndex = 0);
     shared_ptr<Cursor> findTableScan(const char *ns, const BSONObj& order, const DiskLoc &startLoc=DiskLoc());
 
 // -1 if library unavailable.