Merge branch 'master' of git@github.com:mongodb/mongo

38 files changed, 225 insertions, 81 deletions

diff --git a/buildscripts/smoke.py b/buildscripts/smoke.py
index 2568acbe00e..6ae8b32aa97 100755
--- a/buildscripts/smoke.py
+++ b/buildscripts/smoke.py
@@ -145,7 +145,7 @@ class mongod(object):
         utils.ensureDir(dir_name)
         argv = [mongod_executable, "--port", str(self.port), "--dbpath", dir_name]
         if self.kwargs.get('small_oplog'):
-            argv += ["--master", "--oplogSize", "128"]
+            argv += ["--master", "--oplogSize", "256"]
         if self.slave:
             argv += ['--slave', '--source', 'localhost:' + str(srcport)]
         print "running " + " ".join(argv)
diff --git a/client/dbclient.h b/client/dbclient.h
index 5d6d87cc129..03030a3c6be 100644
--- a/client/dbclient.h
+++ b/client/dbclient.h
@@ -884,7 +884,7 @@ namespace mongo {
          */
         bool isFailed() const { return _failed; }
 
-        MessagingPort& port() { return *p; }
+        MessagingPort& port() { assert(p); return *p; }
 
         string toStringLong() const {
             stringstream ss;
diff --git a/client/dbclient_rs.cpp b/client/dbclient_rs.cpp
index 9cf6a18440f..217c38804a5 100644
--- a/client/dbclient_rs.cpp
+++ b/client/dbclient_rs.cpp
@@ -74,7 +74,7 @@ namespace mongo {
 
 
     ReplicaSetMonitor::ReplicaSetMonitor( const string& name , const vector<HostAndPort>& servers )
-        : _lock( "ReplicaSetMonitor instance" ) , _checkConnectionLock( "ReplicaSetMonitor check connection lock" ), _name( name ) , _master(-1) {
+        : _lock( "ReplicaSetMonitor instance" ) , _checkConnectionLock( "ReplicaSetMonitor check connection lock" ), _name( name ) , _master(-1), _nextSlave(0) {
         
         uassert( 13642 , "need at least 1 node for a replica set" , servers.size() > 0 );
 
@@ -221,19 +221,19 @@ namespace mongo {
     }
 
     HostAndPort ReplicaSetMonitor::getSlave() {
-        int x = rand() % _nodes.size();
+
         {
             scoped_lock lk( _lock );
             for ( unsigned i=0; i<_nodes.size(); i++ ) {
-                int p = ( i + x ) % _nodes.size();
-                if ( p == _master )
+                _nextSlave = ( _nextSlave + 1 ) % _nodes.size();
+                if ( _nextSlave == _master )
                     continue;
-                if ( _nodes[p].ok )
-                    return _nodes[p].addr;
+                if ( _nodes[ _nextSlave ].ok )
+                    return _nodes[ _nextSlave ].addr;
             }
         }
 
-        return _nodes[0].addr;
+        return _nodes[ 0 ].addr;
     }
 
     /**
@@ -309,7 +309,7 @@ namespace mongo {
             BSONObj o;
             c->isMaster(isMaster, &o);
 
-            log( ! verbose ) << "ReplicaSetMonitor::_checkConnection: " << c->toString() << ' ' << o << '\n';
+            log( ! verbose ) << "ReplicaSetMonitor::_checkConnection: " << c->toString() << ' ' << o << endl;
 
             // add other nodes
             string maybePrimary;
@@ -530,12 +530,12 @@ namespace mongo {
             // we're ok sending to a slave
             // we'll try 2 slaves before just using master
             // checkSlave will try a different slave automatically after a failure
-            for ( int i=0; i<2; i++ ) {
+            for ( int i=0; i<3; i++ ) {
                 try {
                     return checkSlave()->query(ns,query,nToReturn,nToSkip,fieldsToReturn,queryOptions,batchSize);
                 }
-                catch ( DBException & ) {
-                    LOG(1) << "can't query replica set slave: " << _slaveHost << endl;
+                catch ( DBException &e ) {
+                    LOG(1) << "can't query replica set slave " << i << " : " << _slaveHost << causedBy( e ) << endl;
                 }
             }
         }
@@ -548,12 +548,12 @@ namespace mongo {
             // we're ok sending to a slave
             // we'll try 2 slaves before just using master
             // checkSlave will try a different slave automatically after a failure
-            for ( int i=0; i<2; i++ ) {
+            for ( int i=0; i<3; i++ ) {
                 try {
                     return checkSlave()->findOne(ns,query,fieldsToReturn,queryOptions);
                 }
-                catch ( DBException & ) {
-                    LOG(1) << "can't query replica set slave: " << _slaveHost << endl;
+                catch ( DBException &e ) {
+                	LOG(1) << "can't findone replica set slave " << i << " : " << _slaveHost << causedBy( e ) << endl;
                 }
             }
         }
@@ -581,12 +581,12 @@ namespace mongo {
             DbMessage dm( toSend );
             QueryMessage qm( dm );
             if ( qm.queryOptions & QueryOption_SlaveOk ) {
-                for ( int i=0; i<2; i++ ) {
+                for ( int i=0; i<3; i++ ) {
                     try {
                         return checkSlave()->callLazy( toSend );
                     }
-                    catch ( DBException & ) {
-                        log(1) << "can't query replica set slave: " << _slaveHost << endl;
+                    catch ( DBException &e ) {
+                    	LOG(1) << "can't callLazy replica set slave " << i << " : " << _slaveHost << causedBy( e ) << endl;
                     }
                 }
             }
@@ -601,15 +601,15 @@ namespace mongo {
             DbMessage dm( toSend );
             QueryMessage qm( dm );
             if ( qm.queryOptions & QueryOption_SlaveOk ) {
-                for ( int i=0; i<2; i++ ) {
+                for ( int i=0; i<3; i++ ) {
                     try {
                         DBClientConnection* s = checkSlave();
                         if ( actualServer )
                             *actualServer = s->getServerAddress();
                         return s->call( toSend , response , assertOk );
                     }
-                    catch ( DBException & ) {
-                        log(1) << "can't query replica set slave: " << _slaveHost << endl;
+                    catch ( DBException &e ) {
+                    	LOG(1) << "can't call replica set slave " << i << " : " << _slaveHost << causedBy( e ) << endl;
                         if ( actualServer )
                             *actualServer = "";
                     }
diff --git a/client/dbclient_rs.h b/client/dbclient_rs.h
index f65fe3d3ad4..f0e855ae8c4 100644
--- a/client/dbclient_rs.h
+++ b/client/dbclient_rs.h
@@ -147,7 +147,7 @@ namespace mongo {
         vector<Node> _nodes;
 
         int _master; // which node is the current master.  -1 means no master is known
-
+        int _nextSlave; // which node is the current slave
 
         static mongo::mutex _setsLock; // protects _sets
         static map<string,ReplicaSetMonitorPtr> _sets; // set name to Monitor
diff --git a/client/examples/clientTest.cpp b/client/examples/clientTest.cpp
index 96c014e245c..aaea6bd1bdf 100644
--- a/client/examples/clientTest.cpp
+++ b/client/examples/clientTest.cpp
@@ -246,5 +246,34 @@ int main( int argc, const char **argv ) {
         //MONGO_PRINT(out);
     }
 
+    { 
+        // test timeouts
+
+        DBClientConnection conn( true , 0 , 2 );
+        if ( ! conn.connect( string( "127.0.0.1:" ) + port , errmsg ) ) {
+            cout << "couldn't connect : " << errmsg << endl;
+            throw -11;
+        }
+        conn.insert( "test.totest" , BSON( "x" << 1 ) );
+        BSONObj res;
+        
+        bool gotError = false;
+        assert( conn.eval( "test" , "return db.totest.findOne().x" , res ) );
+        try {
+            conn.eval( "test" , "sleep(5000); return db.totest.findOne().x" , res );
+        }
+        catch ( std::exception& e ) {
+            gotError = true;
+            log() << e.what() << endl;
+        }
+        assert( gotError );
+        // sleep so the server isn't locked anymore
+        sleepsecs( 4 );
+        
+        assert( conn.eval( "test" , "return db.totest.findOne().x" , res ) );
+        
+        
+    }
+
     cout << "client test finished!" << endl;
 }
diff --git a/db/db.vcxproj b/db/db.vcxproj
index a032653c94c..eb56d6fe77f 100644
--- a/db/db.vcxproj
+++ b/db/db.vcxproj
@@ -493,6 +493,10 @@
     <ClCompile Include="mongommf.cpp" />

     <ClCompile Include="oplog.cpp" />

     <ClCompile Include="projection.cpp" />

+    <ClCompile Include="querypattern.cpp">

+      <PrecompiledHeader Condition="'$(Configuration)|$(Platform)'=='Debug|x64'">NotUsing</PrecompiledHeader>

+      <PrecompiledHeader Condition="'$(Configuration)|$(Platform)'=='Release|x64'">NotUsing</PrecompiledHeader>

+    </ClCompile>

     <ClCompile Include="repl.cpp" />

     <ClCompile Include="repl\consensus.cpp" />

     <ClCompile Include="repl\heartbeat.cpp" />

@@ -658,6 +662,7 @@
     <ClInclude Include="namespace-inl.h" />

     <ClInclude Include="oplogreader.h" />

     <ClInclude Include="projection.h" />

+    <ClInclude Include="queryutil.h" />

     <ClInclude Include="repl.h" />

     <ClInclude Include="replpair.h" />

     <ClInclude Include="repl\connections.h" />

diff --git a/db/db.vcxproj.filters b/db/db.vcxproj.filters
index 1ab8438371b..936d0da5d5e 100755
--- a/db/db.vcxproj.filters
+++ b/db/db.vcxproj.filters
@@ -155,6 +155,7 @@
     <ClCompile Include="repl\rs_config.cpp" />

     <ClCompile Include="security_key.cpp" />

     <ClCompile Include="..\util\file_allocator.cpp" />

+    <ClCompile Include="querypattern.cpp" />

   </ItemGroup>

   <ItemGroup>

     <ClInclude Include="..\client\dbclientcursor.h" />

@@ -293,6 +294,7 @@
     <ClInclude Include="dur_journalimpl.h" />

     <ClInclude Include="..\util\concurrency\race.h" />

     <ClInclude Include="..\util\alignedbuilder.h" />

+    <ClInclude Include="queryutil.h" />

   </ItemGroup>

   <ItemGroup>

     <ResourceCompile Include="db.rc" />

diff --git a/db/dur_journal.cpp b/db/dur_journal.cpp
index 52498b900e3..b8c259a6763 100644
--- a/db/dur_journal.cpp
+++ b/db/dur_journal.cpp
@@ -457,7 +457,7 @@ namespace mongo {
             if something highly surprising, throws to abort
         */
         unsigned long long LSNFile::get() {
-            uassert(13614, "unexpected version number of lsn file in journal/ directory", ver == 0);
+            uassert(13614, str::stream() << "unexpected version number of lsn file in journal/ directory got: " << ver , ver == 0);
             if( ~lsn != checkbytes ) {
                 log() << "lsnfile not valid. recovery will be from log start. lsn: " << hex << lsn << " checkbytes: " << hex << checkbytes << endl;
                 return 0;
@@ -486,6 +486,11 @@ namespace mongo {
                 File f;
                 f.open(lsnPath().string().c_str());
                 assert(f.is_open());
+                if( f.len() == 0 ) { 
+                    // this could be 'normal' if we crashed at the right moment
+                    log() << "info lsn file is zero bytes long" << endl;
+                    return 0;
+                }
                 f.read(0,(char*)&L, sizeof(L));
                 unsigned long long lsn = L.get();
                 return lsn;
diff --git a/db/queryutil.h b/db/queryutil.h
index 99a6b874254..a817764246d 100644
--- a/db/queryutil.h
+++ b/db/queryutil.h
@@ -278,7 +278,7 @@ namespace mongo {
         FieldRangeSet _singleKey;
         FieldRangeSet _multiKey;
         friend class FieldRangeOrSet;
-        friend class QueryUtilIndexed;
+        friend struct QueryUtilIndexed;
     };
     
     class IndexSpec;
@@ -407,7 +407,7 @@ namespace mongo {
         // ensure memory is owned
         list<FieldRangeSetPair> _oldOrSets;
         bool _orFound;
-        friend class QueryUtilIndexed;
+        friend struct QueryUtilIndexed;
     };
 
     /** returns a string that when used as a matcher, would match a super set of regex()
diff --git a/db/repl/rs.cpp b/db/repl/rs.cpp
index eea51e3f2e2..84ef3935a3b 100644
--- a/db/repl/rs.cpp
+++ b/db/repl/rs.cpp
@@ -465,9 +465,14 @@ namespace mongo {
         forgetPrimary();
 
         setSelfTo(0);
+
+        // For logging
+        string members = "";
+
         for( vector<ReplSetConfig::MemberCfg>::iterator i = _cfg->members.begin(); i != _cfg->members.end(); i++ ) {
             const ReplSetConfig::MemberCfg& m = *i;
             Member *mi;
+            members += ( members == "" ? "" : ", " ) + m.h.toString();
             if( m.h.isSelf() ) {
                 assert( _self == 0 );
                 mi = new Member(m.h, m._id, &m, true);
@@ -484,6 +489,11 @@ namespace mongo {
                     box.setOtherPrimary(mi);
             }
         }
+
+        if( ! _self ){
+            log() << "replSet warning did not detect own host in full reconfig, members " << members << " config: " << c << rsLog;
+        }
+
         return true;
     }
 
diff --git a/db/repl/rs.h b/db/repl/rs.h
index 9c1948e67a4..5b8863aa161 100644
--- a/db/repl/rs.h
+++ b/db/repl/rs.h
@@ -379,7 +379,7 @@ namespace mongo {
         void loadConfig();
 
         list<HostAndPort> memberHostnames() const;
-        const ReplSetConfig::MemberCfg& myConfig() const { return _self->config(); }
+        const ReplSetConfig::MemberCfg& myConfig() const { assert( _self ); return _self->config(); }
         bool iAmArbiterOnly() const { return myConfig().arbiterOnly; }
         bool iAmPotentiallyHot() const { return myConfig().potentiallyHot() && elect.steppedDown <= time(0); }
     protected:
@@ -505,6 +505,7 @@ namespace mongo {
 
     inline Member::Member(HostAndPort h, unsigned ord, const ReplSetConfig::MemberCfg *c, bool self) :
         _config(*c), _h(h), _hbinfo(ord) {
+        assert(c);
         if( self )
             _hbinfo.health = 1.0;
     }
diff --git a/db/repl/rs_config.cpp b/db/repl/rs_config.cpp
index 28b26154836..229bddc120e 100644
--- a/db/repl/rs_config.cpp
+++ b/db/repl/rs_config.cpp
@@ -68,7 +68,7 @@ namespace mongo {
 
             cx.db()->flushFiles(true);
         }
-        DEV log() << "replSet saveConfigLocally done" << rsLog;
+        log() << "replSet saveConfigLocally done" << rsLog;
     }
 
     bo ReplSetConfig::MemberCfg::asBson() const {
diff --git a/db/repl/rs_sync.cpp b/db/repl/rs_sync.cpp
index d14581fab4a..535796b5a25 100644
--- a/db/repl/rs_sync.cpp
+++ b/db/repl/rs_sync.cpp
@@ -34,7 +34,7 @@ namespace mongo {
         if ( *ns == '.' || *ns == 0 ) {
             if( *o.getStringField("op") == 'n' )
                 return;
-            log() << "replSet skipping bad op in oplog: " << o.toString() << endl;
+            log() << "replSet skipping bad op in oplog: " << o.toString() << rsLog;
             return;
         }
 
@@ -141,7 +141,7 @@ namespace mongo {
                     time_t now = time(0);
                     if (now - start > 10) {
                         // simple progress metering
-                        log() << "initialSyncOplogApplication applied " << n << " operations, synced to "
+                        log() << "replSet initialSyncOplogApplication applied " << n << " operations, synced to "
                               << ts.toStringPretty() << rsLog;
                         start = now;
                     }
@@ -322,8 +322,8 @@ namespace mongo {
             OpTime ts = o["ts"]._opTime();
             long long h = o["h"].numberLong();
             if( ts != lastOpTimeWritten || h != lastH ) {
-                log() << "replSet our last op time written: " << lastOpTimeWritten.toStringPretty() << endl;
-                log() << "replset source's GTE: " << ts.toStringPretty() << endl;
+                log() << "replSet our last op time written: " << lastOpTimeWritten.toStringPretty() << rsLog;
+                log() << "replset source's GTE: " << ts.toStringPretty() << rsLog;
                 syncRollback(r);
                 return;
             }
@@ -467,8 +467,13 @@ namespace mongo {
         */
 
         while( 1 ) {
-            if( myConfig().arbiterOnly )
+            // After a reconfig, we may not be in the replica set anymore, so
+            // check that we are in the set (and not an arbiter) before
+            // trying to sync with other replicas.
+            if( ! _self || myConfig().arbiterOnly ){
+            	if( ! _self ) log() << "replSet warning did not detect own host and port, not syncing, config: " << theReplSet->config() << rsLog;
                 return;
+            }
 
             try {
                 _syncThread();
@@ -488,7 +493,11 @@ namespace mongo {
                are no heartbeat threads, so we do it here to be sure.  this is relevant if the singleton
                member has done a stepDown() and needs to come back up.
                */
-            OCCASIONALLY mgr->send( boost::bind(&Manager::msgCheckNewState, theReplSet->mgr) );
+            OCCASIONALLY {
+            	log() << "replSet default heartbeat starting..." << rsLog;
+            	mgr->send( boost::bind(&Manager::msgCheckNewState, theReplSet->mgr) );
+            	log() << "replSet heartbeat finished" << rsLog;
+            }
         }
     }
 
diff --git a/dbtests/test.vcxproj b/dbtests/test.vcxproj
index 984272d6429..66c890e9d44 100644
--- a/dbtests/test.vcxproj
+++ b/dbtests/test.vcxproj
@@ -285,6 +285,10 @@
     <ClCompile Include="..\db\geo\haystack.cpp" />

     <ClCompile Include="..\db\mongommf.cpp" />

     <ClCompile Include="..\db\projection.cpp" />

+    <ClCompile Include="..\db\querypattern.cpp">

+      <PrecompiledHeader Condition="'$(Configuration)|$(Platform)'=='Debug|x64'">NotUsing</PrecompiledHeader>

+      <PrecompiledHeader Condition="'$(Configuration)|$(Platform)'=='Release|x64'">NotUsing</PrecompiledHeader>

+    </ClCompile>

     <ClCompile Include="..\db\repl\consensus.cpp" />

     <ClCompile Include="..\db\repl\heartbeat.cpp" />

     <ClCompile Include="..\db\repl\manager.cpp" />

diff --git a/dbtests/test.vcxproj.filters b/dbtests/test.vcxproj.filters
index c576e5386e4..15d2c98d329 100755
--- a/dbtests/test.vcxproj.filters
+++ b/dbtests/test.vcxproj.filters
@@ -773,6 +773,9 @@
     <ClCompile Include="..\db\dbcommands_admin.cpp">

       <Filter>db\cpp</Filter>

     </ClCompile>

+    <ClCompile Include="..\db\querypattern.cpp">

+      <Filter>db</Filter>

+    </ClCompile>

   </ItemGroup>

   <ItemGroup>

     <None Include="..\SConstruct">

diff --git a/jstests/check_shard_index.js b/jstests/check_shard_index.js
index a5a1fc14c47..e63b79ddc23 100644
--- a/jstests/check_shard_index.js
+++ b/jstests/check_shard_index.js
@@ -42,4 +42,9 @@ assert.eq( 3 , f.count() , "2. count after initial insert should be 3" );
 res = db.runCommand( { checkShardingIndex: "test.jstests_shardingindex" , keyPattern: {x:1, y:1} , force: true });
 assert.eq( false , res.ok , "2b " + tojson(res) );
 
+// 
+res = db.runCommand( { checkShardingIndex: "test.jstests_shardingindex" , keyPattern: {_id:1} , force: true });
+assert.eq( true , res.ok , "3a " + tojson(res) );
+assert( res.idskip , "3b " + tojson(res) )
+
 print("PASSED");
diff --git a/jstests/delx.js b/jstests/delx.js
index 3f8c88c1648..aa858e92cbd 100644
--- a/jstests/delx.js
+++ b/jstests/delx.js
@@ -23,6 +23,7 @@ x.next();
 y.next();
 
 a.foo.remove( { _id : { $gt : 50 } } );
+db.getLastError();
 
 assert.eq( 51 , a.foo.find().itcount() , "B1" )
 assert.eq( 100 , b.foo.find().itcount() , "B2" )
diff --git a/jstests/drop3.js b/jstests/drop3.js
index 1badf07d54b..b2ca94a1550 100644
--- a/jstests/drop3.js
+++ b/jstests/drop3.js
@@ -19,6 +19,7 @@ assert.eq( cursor.objsLeftInBatch(), 1 );
 assert.eq( subcursor.objsLeftInBatch(), 1 );
 
 t.drop(); // should invalidate cursor, but not subcursor
+db.getLastError();
 
 assert.throws( function(){ cursor.itcount() } );  // throws "cursor doesn't exist on server" error on getMore
 assert.eq( subcursor.itcount(), 9 ); //one already seen
diff --git a/jstests/dropdb.js b/jstests/dropdb.js
index 424cfca2b38..0b838846bde 100644
--- a/jstests/dropdb.js
+++ b/jstests/dropdb.js
@@ -7,6 +7,7 @@ baseName = "jstests_dropdb";
 ddb = db.getSisterDB( baseName );
 
 ddb.c.save( {} );
+ddb.getLastError();
 assert.neq( -1, m.getDBNames().indexOf( baseName ) );
 
 ddb.dropDatabase();
diff --git a/jstests/evalc.js b/jstests/evalc.js
index 8a9e889c6ef..0320ecd5133 100644
--- a/jstests/evalc.js
+++ b/jstests/evalc.js
@@ -1,17 +1,24 @@
 t = db.jstests_evalc;
 t.drop();
 
+t2 = db.evalc_done
+t2.drop()
+
 for( i = 0; i < 10; ++i ) {
     t.save( {i:i} );
 }
 
 // SERVER-1610
 
-s = startParallelShell( "print( 'starting forked:' + Date() ); for ( i=0; i<500000; i++ ){ db.currentOp(); } print( 'ending forked:' + Date() ); " )
+assert.eq( 0 , t2.count() , "X1" )
+
+s = startParallelShell( "print( 'starting forked:' + Date() ); for ( i=0; i<50000; i++ ){ db.currentOp(); } print( 'ending forked:' + Date() ); db.evalc_done.insert( { x : 1 } ); " )
 
 print( "starting eval: " + Date() )
-for ( i=0; i<20000; i++ ){
+while ( true ) {
     db.eval( "db.jstests_evalc.count( {i:10} );" );
+    if ( t2.count() > 0 )
+        break;
 }
 print( "end eval: " + Date() )
 
diff --git a/jstests/indexs.js b/jstests/indexs.js
new file mode 100644
index 00000000000..3a52584bfd3
--- /dev/null
+++ b/jstests/indexs.js
@@ -0,0 +1,23 @@
+// Test index key generation issue with parent and nested fields in same index and array containing subobject SERVER-3005.
+
+t = db.jstests_indexs;
+
+t.drop();
+t.ensureIndex( {a:1} );
+t.save( { a: [ { b: 3 } ] } );
+assert.eq( 1, t.count( { a:{ b:3 } } ) );
+
+t.drop();
+t.ensureIndex( {a:1,'a.b':1} );
+t.save( { a: { b: 3 } } );
+assert.eq( 1, t.count( { a:{ b:3 } } ) );
+ib = t.find( { a:{ b:3 } } ).explain().indexBounds;
+
+t.drop();
+t.ensureIndex( {a:1,'a.b':1} );
+t.save( { a: [ { b: 3 } ] } );
+assert.eq( ib, t.find( { a:{ b:3 } } ).explain().indexBounds );
+if ( 0 ) { // SERVER-3005
+assert.eq( 1, t.find( { a:{ b:3 } } ).explain().nscanned );
+assert.eq( 1, t.count( { a:{ b:3 } } ) );
+}
\ No newline at end of file
diff --git a/jstests/or5.js b/jstests/or5.js
index baa6bd6f847..98ff141f575 100644
--- a/jstests/or5.js
+++ b/jstests/or5.js
@@ -70,6 +70,7 @@ assert.eq.automsg( "6", "t.find( {$or:[{a:2},{b:3},{c:4}]} ).batchSize( 2 ).itco
 c = t.find( {$or:[{a:2},{b:3},{c:4}]} ).batchSize( 2 );
 c.next();
 t.remove( {b:3} );
+db.getLastError();
 assert.eq.automsg( "3", c.itcount() );
 
 reset();
@@ -78,6 +79,7 @@ c = t.find( {$or:[{a:2},{b:3},{c:4}]} ).batchSize( 2 );
 c.next();
 c.next();
 t.remove( {b:3} );
+db.getLastError();
 assert.eq.automsg( "2", c.itcount() );
 
 reset();
@@ -87,6 +89,7 @@ c.next();
 c.next();
 c.next();
 t.remove( {b:3} );
+db.getLastError();
 assert.eq.automsg( "3", c.itcount() );
 
 reset();
@@ -97,6 +100,7 @@ c.next();
 c.next();
 c.next();
 t.remove( {b:3} );
+db.getLastError();
 assert.eq.automsg( "2", c.itcount() );
 
 t.drop();
diff --git a/jstests/ord.js b/jstests/ord.js
index 4612f218f64..f78e5044074 100644
--- a/jstests/ord.js
+++ b/jstests/ord.js
@@ -28,6 +28,7 @@ for( i = 0; i < 90; ++i ) {
 // the index key {a:1}
 
 t.dropIndex( {a:1} );
+db.getLastError();
 
 // Dropping an index kills all cursors on the indexed namespace, not just those 
 // cursors using the dropped index.
diff --git a/jstests/replsets/remove1.js b/jstests/replsets/remove1.js
index b5f9a4f1a00..d6d36e40f25 100644
--- a/jstests/replsets/remove1.js
+++ b/jstests/replsets/remove1.js
@@ -82,8 +82,9 @@ wait(function() {
     printjson(master.getDB("admin").runCommand({replSetGetStatus:1}));
     master.setSlaveOk();
     var newConfig = master.getDB("local").system.replset.findOne();
+    print( "newConfig: " + tojson(newConfig) );
     return newConfig.version == 4;
-  });
+} , "wait1" );
 
 
 print("Make sure everyone's secondary");
@@ -103,7 +104,7 @@ wait(function() {
       }
     }
     return true;
-  });
+} , "wait2" );
 
 replTest.stopSet();
 
diff --git a/jstests/replsets/rslib.js b/jstests/replsets/rslib.js
index 6434cddcf96..19271c9e13d 100644
--- a/jstests/replsets/rslib.js
+++ b/jstests/replsets/rslib.js
@@ -2,7 +2,7 @@
 var count = 0;
 var w = 0;
 
-var wait = function(f) {
+var wait = function(f,msg) {
     w++;
     var n = 0;
     while (!f()) {
@@ -11,7 +11,7 @@ var wait = function(f) {
         if (++n == 4) {
             print("" + f);
         }
-        assert(n < 200, 'tried 200 times, giving up');
+        assert(n < 200, 'tried 200 times, giving up on ' + msg );
         sleep(1000);
     }
 };
diff --git a/jstests/replsets/sync1.js b/jstests/replsets/sync1.js
index de824d0989b..5e862106efa 100644
--- a/jstests/replsets/sync1.js
+++ b/jstests/replsets/sync1.js
@@ -80,12 +80,17 @@ doTest = function (signal) {
 
     assert(count < 300);
     
+    // Need to be careful here, allocating datafiles for the slaves can take a *long* time on slow systems
+    sleep(7000);
+
     print("\nsync1.js ********************************************************************** part 6");
     dbs[0].getSisterDB("admin").runCommand({ replSetTest: 1, blind: true });
 
     print("\nsync1.js ********************************************************************** part 7");
 
     sleep(5000);
+    // If we start getting error hasNext: false with done alloc datafile msgs - may need to up the sleep again in part 5
+
 
     var max1;
     var max2;
diff --git a/jstests/sharding/migrateBig.js b/jstests/sharding/migrateBig.js
index f6ba18a7c02..917f152a4d2 100644
--- a/jstests/sharding/migrateBig.js
+++ b/jstests/sharding/migrateBig.js
@@ -40,6 +40,6 @@ for ( i=0; i<20; i+= 2 )
 
 db.printShardingStatus()
 
-assert.soon( function(){ var x = s.chunkDiff( "foo" , "test" ); print( "chunk diff: " + x ); return x < 2; } , "no balance happened" , 120 * 1000 , 2000 ) 
+assert.soon( function(){ var x = s.chunkDiff( "foo" , "test" ); print( "chunk diff: " + x ); return x < 2; } , "no balance happened" , 8 * 60 * 1000 , 2000 ) 
 
 s.stop()
diff --git a/jstests/sharding/shard_insert_getlasterror_w2.js b/jstests/sharding/shard_insert_getlasterror_w2.js
index c722f216e64..1e0266dcc7d 100644
--- a/jstests/sharding/shard_insert_getlasterror_w2.js
+++ b/jstests/sharding/shard_insert_getlasterror_w2.js
@@ -75,7 +75,8 @@ function go() {
                 return false;
             }
             return true;
-        });
+        }, "Queries took too long to complete correctly.",
+        2 * 60 * 1000 );
     
     // Done
     routerSpec.end()
diff --git a/jstests/sharding/sync7.js b/jstests/sharding/sync7.js
index 0badd9ebeb7..a8ff0944a00 100644
--- a/jstests/sharding/sync7.js
+++ b/jstests/sharding/sync7.js
@@ -7,7 +7,7 @@ s._connections[1].getDB( "admin" ).runCommand( { _skewClockCommand : 1, skew : -
 
 // We need to start another mongos after skewing the clock, since the first mongos will have already
 // tested the config servers (via the balancer) before we manually skewed them
-otherMongos = startMongos( { port : 30020, v : 0, configdb : s._configDB, logpath : "/dev/null" } );
+otherMongos = startMongos( { port : 30020, v : 0, configdb : s._configDB } );
 
 // Initialize DB data
 initDB = function(name) {
diff --git a/s/balance.cpp b/s/balance.cpp
index 40d542b9111..ffa23ed794b 100644
--- a/s/balance.cpp
+++ b/s/balance.cpp
@@ -275,20 +275,21 @@ namespace mongo {
 
             try {
                 
-                // first make sure we should even be running
+                ScopedDbConnection conn( config );
+                
+                // ping has to be first so we keep things in the config server in sync
+                _ping( conn.conn() );
+
+                // now make sure we should even be running
                 if ( ! grid.shouldBalance() ) {
                     log(1) << "skipping balancing round because balancing is disabled" << endl;
+                    conn.done();
+                    
                     sleepsecs( 30 );
                     continue;
                 }
                 
-
-                ScopedDbConnection conn( config );
-
-                _ping( conn.conn() );
-                if ( ! _checkOIDs() ) {
-                    uassert( 13258 , "oids broken after resetting!" , _checkOIDs() );
-                }
+                uassert( 13258 , "oids broken after resetting!" , _checkOIDs() );
 
                 // use fresh shard state
                 Shard::reloadShardInfo();
diff --git a/s/chunk.cpp b/s/chunk.cpp
index 0fb2fca1237..e63c38bae44 100644
--- a/s/chunk.cpp
+++ b/s/chunk.cpp
@@ -726,7 +726,6 @@ namespace mongo {
 
     void ChunkManager::getShardsForQuery( set<Shard>& shards , const BSONObj& query ) {
         rwlock lk( _lock , false );
-        DEV PRINT(query);
 
         //TODO look into FieldRangeSetOr
         FieldRangeOrSet fros(_ns.c_str(), query, false);
@@ -762,9 +761,6 @@ namespace mongo {
                 BSONObj minObj = it->first.replaceFieldNames(_key.key());
                 BSONObj maxObj = it->second.replaceFieldNames(_key.key());
 
-                DEV PRINT(minObj);
-                DEV PRINT(maxObj);
-
                 ChunkRangeMap::const_iterator min, max;
                 min = _chunkRanges.upper_bound(minObj);
                 max = _chunkRanges.upper_bound(maxObj);
diff --git a/s/d_migrate.cpp b/s/d_migrate.cpp
index fe4c0efe355..22b600a8db1 100644
--- a/s/d_migrate.cpp
+++ b/s/d_migrate.cpp
@@ -1008,6 +1008,7 @@ namespace mongo {
                     conn.done();
                 }
                 catch ( DBException& e ) {
+                    warning() << e << endl;
                     ok = false;
                     BSONObjBuilder b;
                     e.getInfo().append( b );
diff --git a/s/d_split.cpp b/s/d_split.cpp
index 59a07420956..6eaf4cf502e 100644
--- a/s/d_split.cpp
+++ b/s/d_split.cpp
@@ -138,6 +138,11 @@ namespace mongo {
             const char* ns = jsobj.getStringField( "checkShardingIndex" );
             BSONObj keyPattern = jsobj.getObjectField( "keyPattern" );
 
+            if ( keyPattern.nFields() == 1 && str::equals( "_id" , keyPattern.firstElement().fieldName() ) ) {
+                result.appendBool( "idskip" , true );
+                return true;
+            }
+
             // If min and max are not provided use the "minKey" and "maxKey" for the sharding key pattern.
             BSONObj min = jsobj.getObjectField( "min" );
             BSONObj max = jsobj.getObjectField( "max" );
@@ -211,6 +216,9 @@ namespace mongo {
                     return false;
                 }
                 cc->advance();
+
+                if ( ! cc->yieldSometimes() ) 
+                    break;
             }
 
             return true;
diff --git a/s/dbgrid.vcxproj b/s/dbgrid.vcxproj
index 9affd0e0b6c..507d2126b1a 100644
--- a/s/dbgrid.vcxproj
+++ b/s/dbgrid.vcxproj
@@ -193,6 +193,10 @@
     <ClCompile Include="..\client\distlock.cpp" />

     <ClCompile Include="..\db\dbcommands_generic.cpp" />

     <ClCompile Include="..\db\dbwebserver.cpp" />

+    <ClCompile Include="..\db\querypattern.cpp">

+      <PrecompiledHeader Condition="'$(Configuration)|$(Platform)'=='Debug|x64'">NotUsing</PrecompiledHeader>

+      <PrecompiledHeader Condition="'$(Configuration)|$(Platform)'=='Release|x64'">NotUsing</PrecompiledHeader>

+    </ClCompile>

     <ClCompile Include="..\db\security_key.cpp" />

     <ClCompile Include="..\scripting\bench.cpp" />

     <ClCompile Include="..\util\alignedbuilder.cpp">

diff --git a/s/dbgrid.vcxproj.filters b/s/dbgrid.vcxproj.filters
index 1fa39311400..46c441feba6 100755
--- a/s/dbgrid.vcxproj.filters
+++ b/s/dbgrid.vcxproj.filters
@@ -323,6 +323,9 @@
     <ClCompile Include="..\db\dbcommands_generic.cpp">

       <Filter>Shared Source Files</Filter>

     </ClCompile>

+    <ClCompile Include="..\db\querypattern.cpp">

+      <Filter>Shared Source Files</Filter>

+    </ClCompile>

   </ItemGroup>

   <ItemGroup>

     <ClInclude Include="gridconfig.h">

diff --git a/tools/sniffer.cpp b/tools/sniffer.cpp
index 4e9cfd21bf3..024a37475cc 100644
--- a/tools/sniffer.cpp
+++ b/tools/sniffer.cpp
@@ -104,6 +104,9 @@ struct sniff_ip {
 #define IP_V(ip)                (((ip)->ip_vhl) >> 4)
 
 /* TCP header */
+#ifdef _WIN32
+typedef unsigned __int32 uint32_t;
+#endif
 typedef uint32_t tcp_seq;
 
 struct sniff_tcp {
diff --git a/util/file.h b/util/file.h
index f61c2c7f645..704dc536211 100644
--- a/util/file.h
+++ b/util/file.h
@@ -38,6 +38,8 @@ namespace mongo {
     typedef boost::uint64_t fileofs;
 #endif
 
+    /* NOTE: not thread-safe. (at least the windows implementation isn't. */
+
     class FileInterface {
     public:
         void open(const char *fn) {}
@@ -171,12 +173,23 @@ namespace mongo {
             err( ::pwrite(fd, data, len, o) == (int) len );
         }
         void read(fileofs o, char *data, unsigned len) {
-            err( ::pread(fd, data, len, o) == (int) len );
+            ssize_t s = ::pread(fd, data, len, o);
+            if( s == -1 ) {
+                err(false);
+            }
+            else if( s != (int) len ) { 
+                _bad = true;
+                log() << "File error read:" << s << " bytes, wanted:" << len << " ofs:" << o << endl;
+            }
         }
         bool bad() { return _bad; }
         bool is_open() { return fd > 0; }
         fileofs len() {
-            return lseek(fd, 0, SEEK_END);
+            off_t o = lseek(fd, 0, SEEK_END);
+            if( o != (off_t) -1 )
+                return o;
+            err(false);
+            return 0;
         }
         void fsync() { ::fsync(fd); }
         static boost::intmax_t freeSpace ( const string &path ) {
diff --git a/util/message.cpp b/util/message.cpp
index c58d0ddca63..892d64c7bdb 100644
--- a/util/message.cpp
+++ b/util/message.cpp
@@ -364,7 +364,7 @@ namespace mongo {
         ConnectBG(int sock, SockAddr farEnd) : _sock(sock), _farEnd(farEnd) { }
 
         void run() { _res = ::connect(_sock, _farEnd.raw(), _farEnd.addressSize); }
-        string name() const { return ""; /* too short lived to need to name */ }
+        string name() const { return "ConnectBG"; }
         int inError() const { return _res; }
 
     private:
@@ -634,12 +634,20 @@ again:
         unsigned retries = 0;
         while( len > 0 ) {
             int ret = ::recv( sock , buf , len , portRecvFlags );
-            if ( ret == 0 ) {
+            if ( ret > 0 ) {
+                if ( len <= 4 && ret != len )
+                    log(_logLevel) << "MessagingPort recv() got " << ret << " bytes wanted len=" << len << endl;
+                assert( ret <= len );
+                len -= ret;
+                buf += ret;
+            }
+            else if ( ret == 0 ) {
                 log(3) << "MessagingPort recv() conn closed? " << farEnd.toString() << endl;
                 throw SocketException( SocketException::CLOSED );
             }
-            if ( ret < 0 ) {
+            else { /* ret < 0  */
                 int e = errno;
+                
 #if defined(EINTR) && !defined(_WIN32)
                 if( e == EINTR ) {
                     if( ++retries == 1 ) {
@@ -648,29 +656,18 @@ again:
                     }
                 }
 #endif
-                if ( e != EAGAIN || _timeout == 0 ) {
-                    SocketException::Type t = SocketException::RECV_ERROR;
-#if defined(_WINDOWS)
-                    if( e == WSAETIMEDOUT ) t = SocketException::RECV_TIMEOUT;
-#else
-                    /* todo: what is the error code on an SO_RCVTIMEO on linux? EGAIN? EWOULDBLOCK? */
+                if ( ( e == EAGAIN 
+#ifdef _WINDOWS
+                       || e == WSAETIMEDOUT
 #endif
-                    log(_logLevel) << "MessagingPort recv() " << errnoWithDescription(e) << " " << farEnd.toString() <<endl;
-                    throw SocketException(t);
-                }
-                else {
-                    if ( !serverAlive( farEnd.toString() ) ) {
-                        log(_logLevel) << "MessagingPort recv() remote dead " << farEnd.toString() << endl;
-                        throw SocketException( SocketException::RECV_ERROR );
-                    }
+                       ) && _timeout > 0 ) {
+                    // this is a timeout
+                    log(_logLevel) << "MessagingPort recv() timeout  " << farEnd.toString() <<endl;
+                    throw SocketException(SocketException::RECV_TIMEOUT);                    
                 }
-            }
-            else {
-                if ( len <= 4 && ret != len )
-                    log(_logLevel) << "MessagingPort recv() got " << ret << " bytes wanted len=" << len << endl;
-                assert( ret <= len );
-                len -= ret;
-                buf += ret;
+
+                log(_logLevel) << "MessagingPort recv() " << errnoWithDescription(e) << " " << farEnd.toString() <<endl;
+                throw SocketException(SocketException::RECV_ERROR);
             }
         }
     }