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

bson/bsonobj.h

@@ -227,6 +227,11 @@ namespace mongo {
                       bool considerFieldName=true) const;
 
 
+        bool operator<( const BSONObj& other ) const { return woCompare( other ) < 0; }
+        bool operator<=( const BSONObj& other ) const { return woCompare( other ) <= 0; }
+        bool operator>( const BSONObj& other ) const { return woCompare( other ) > 0; }
+        bool operator>=( const BSONObj& other ) const { return woCompare( other ) >= 0; }
+
         /**
          * @param useDotted whether to treat sort key fields as possibly dotted and expand into them
          */

client/distlock.cpp

@@ -100,7 +100,7 @@ namespace mongo {
 
     } distLockPinger;
 
-    DistributedLock::DistributedLock( const ConnectionString& conn , const string& name , int takeoverMinutes )
+    DistributedLock::DistributedLock( const ConnectionString& conn , const string& name , unsigned takeoverMinutes )
         : _conn(conn),_name(name),_takeoverMinutes(takeoverMinutes){
         _id = BSON( "_id" << name );
         _ns = "config.locks";

client/distlock.h

@@ -34,7 +34,7 @@ namespace mongo {
         /**
          * @param takeoverMinutes how long before we steal lock in minutes
          */
-        DistributedLock( const ConnectionString& conn , const string& name , int takeoverMinutes = 10 );
+        DistributedLock( const ConnectionString& conn , const string& name , unsigned takeoverMinutes = 10 );
 
         int getState(){
             return _state.get();
@@ -50,7 +50,7 @@ namespace mongo {
     private:
         ConnectionString _conn;
         string _name;
-        int _takeoverMinutes;
+        unsigned _takeoverMinutes;
 
         string _ns;
         BSONObj _id;

db/repl/health.cpp

@@ -258,12 +258,12 @@ namespace mongo {
             /* self row */
             s << tr() << td(_self->fullName() + " (me)") <<
                 td(_self->id()) <<
-                td("1") << 
+  	        td("1") <<  //up
                 td(ago(started)) << 
-                td("") << 
+	        td("") << // last heartbeat
                 td(ToString(_self->config().votes)) << 
                 td(stateAsHtml(_myState));
-            s << td( _self->lhb() );
+            s << td( _hbmsg );
             stringstream q;
             q << "/_replSetOplog?" << _self->id();
             s << td( a(q.str(), "", theReplSet->lastOpTimeWritten.toString()) );

db/repl/rs_rollback.cpp

@@ -0,0 +1,69 @@
+/* @file rs_rollback.cpp
+* 
+*    Copyright (C) 2008 10gen Inc.
+*
+*    This program is free software: you can redistribute it and/or  modify
+*    it under the terms of the GNU Affero General Public License, version 3,
+*    as published by the Free Software Foundation.
+*
+*    This program is distributed in the hope that it will be useful,
+*    but WITHOUT ANY WARRANTY; without even the implied warranty of
+*    MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+*    GNU Affero General Public License for more details.
+*
+*    You should have received a copy of the GNU Affero General Public License
+*    along with this program.  If not, see <http://www.gnu.org/licenses/>.
+*/
+
+#include "pch.h"
+#include "../client.h"
+#include "../../client/dbclient.h"
+#include "rs.h"
+#include "../repl.h"
+
+/* Scenarios
+
+   We went offline with ops not replicated out.
+ 
+       F = node that failed and coming back.
+       P = node that took over, new primary
+
+   #1:
+       F : a b c d e f g
+       P : a b c d q
+
+   The design is "keep P".  One could argue here that "keep F" has some merits, however, in most cases P 
+   will have significantly more data.  Also note that P may have a proper subset of F's stream if there were 
+   no subsequent writes!
+
+   For now the model is simply : get F back in sync with P.  If P was really behind or something, we should have 
+   just chosen not to fail over anyway.
+
+   #2:
+       F : a b c d e f g                -> a b c d
+       P : a b c d
+
+   #3:
+       F : a b c d e f g                -> a b c d q r s t u v w x z
+       P : a b c d.q r s t u v w x z
+
+   Steps
+    find an event in common. 'd'.
+    undo our events beyond that by: 
+      (1) taking copy from other server of those objects
+      (2) do not consider copy valid until we pass the original end point (e.g. g) in time
+          -- i.e., reset minvalid.
+      (3) skip operations on objects that are previous in time to our capture of the object.
+
+a b c d e f
+a b c
+a b c d e.g 
+
+*/
+
+namespace mongo {
+
+  void ReplSetImpl::syncRollback(OplogReader&r, ...) { 
+  }
+
+}

db/repl/rs_sync.cpp

@@ -69,11 +69,11 @@ namespace mongo {
             long long h = o["h"].numberLong();
             if( ts != lastOpTimeWritten || h != lastH ) { 
                 if( lastOpTimeWritten < ts ) { 
-                    log() << "replSet ERROR too stale to catch up, at least from primary " << hn << rsLog;
+                    log() << "replSet error too stale to catch up, at least from primary " << hn << rsLog;
                     log() << "replSet our last optime : " << lastOpTimeWritten.toStringPretty() << rsLog;
                     log() << "replSet oldest at " << hn << " : " << ts.toStringPretty() << rsLog;
                     log() << "replSet See http://www.mongodb.org/display/DOCS/Resyncing+a+Very+Stale+Replica+Set+Member" << rsLog;
-                    sethbmsg("sync exception too stale to catch up");
+                    sethbmsg("error too stale to catch up");
                     sleepsecs(120);
                     return;
                 }

dbtests/jsobjtests.cpp

@@ -1606,6 +1606,27 @@ namespace JsobjTests {
         }
     };
 
+    class CompareOps {
+    public:
+        void run(){
+
+            BSONObj a = BSON("a"<<1);
+            BSONObj b = BSON("a"<<1);
+            BSONObj c = BSON("a"<<2);
+            BSONObj d = BSON("a"<<3);
+            BSONObj e = BSON("a"<<4);
+            BSONObj f = BSON("a"<<4);
+
+            ASSERT( ! ( a < b ) );
+            ASSERT( a <= b );
+            ASSERT( a < c );
+
+            ASSERT( f > d );
+            ASSERT( f >= e );
+            ASSERT( ! ( f > e ) );
+        }
+    };
+
     class All : public Suite {
     public:
         All() : Suite( "jsobj" ){
@@ -1712,6 +1733,7 @@ namespace JsobjTests {
             add< BSONFieldTests >();
             add< BSONForEachTest >();
             add< StringDataTest >();
+            add< CompareOps >();
         }
     } myall;
 

jstests/repl/basic1.js

@@ -126,6 +126,17 @@ assert.eq( { _id : "fun" , a : { b : { c : { x : 6848 , y : 911 } } } } , as.b.f
 check( "b 4" );
 
 
+// lots of indexes
+
+am.lotOfIndexes.insert( { x : 1 } )
+for ( i=0; i<200; i++ ){
+    var idx = {}
+    idx["x"+i] = 1;
+    am.lotOfIndexes.ensureIndex( idx );
+}
+
+assert.eq( am.lotOfIndexes.getIndexes().length , as.lotOfIndexes.getIndexes().length , "lots of indexes" )
+
 
 rt.stop();
 

s/chunk.cpp

@@ -120,19 +120,23 @@ namespace mongo {
                 return _manager->getShardKey().extractKey( end );
         }
 
+        BSONObj cmd = BSON( "medianKey" << _manager->getns()
+                            << "keyPattern" << _manager->getShardKey().key()
+                            << "min" << getMin()
+                            << "max" << getMax() );
+
         ScopedDbConnection conn( getShard().getConnString() );
         BSONObj result;
-        if ( ! conn->runCommand( "admin" , BSON( "medianKey" << _manager->getns()
-                                                 << "keyPattern" << _manager->getShardKey().key()
-                                                 << "min" << getMin()
-                                                 << "max" << getMax()
-                                                 ) , result ) ){
+        if ( ! conn->runCommand( "admin" , cmd , result ) ){
             stringstream ss;
             ss << "medianKey command failed: " << result;
             uassert( 10164 ,  ss.str() , 0 );
         }
 
-        BSONObj median = result.getObjectField( "median" );
+        BSONObj median = result.getObjectField( "median" ).getOwned();
+        conn.done();
+
+
         if (median == getMin()){
             Query q;
             q.minKey(_min).maxKey(_max);
@@ -141,10 +145,16 @@ namespace mongo {
             median = conn->findOne(_manager->getns(), q);
             median = _manager->getShardKey().extractKey( median );
         }
-
-        conn.done();
 
-        return median.getOwned();
+        if ( median < getMin() || median >= getMax() ){
+            stringstream ss;
+            ss << "medianKey returned value out of range.  " 
+               << " cmd: " << cmd 
+               << " result: " << result;
+            uasserted( 13394 , ss.str() );
+        }
+
+        return median;
     }
 
     void Chunk::pickSplitVector( vector<BSONObj>* splitPoints ) const { 
@@ -215,6 +225,14 @@ namespace mongo {
             log(4) << "splitPoint: " << splitPoint << endl;
             nextPoint = (++i != m.end()) ? i->getOwned() : _max.getOwned();
             log(4) << "nextPoint: " << nextPoint << endl;
+
+            if ( nextPoint <= splitPoint) {
+                stringstream ss;
+                ss << "multiSplit failing because keys min: " << splitPoint << " and max: " << nextPoint
+                   << " do not define a valid chunk";
+                uasserted( 13395, ss.str() );
+            }
+
             ChunkPtr s( new Chunk( _manager, splitPoint , nextPoint , _shard) );
             s->_markModified();
             newChunks.push_back(s);
@@ -349,7 +367,7 @@ namespace mongo {
         if ( size < myMax )
             return false;
 
-        log() << "autosplitting " << _manager->getns() << " size: " << size << " shard: " << toString() << endl;
+        log() << "autosplitting " << _manager->getns() << " size: " << size << " shard: " << toString() << " on: " << splitPoint << endl;
         vector<BSONObj> splitPoints;
         splitPoints.push_back( splitPoint );
         ChunkPtr newShard = multiSplit( splitPoints );

s/d_split.cpp

@@ -80,7 +80,25 @@ namespace mongo {
                 return false;
             }
 
-            result.append( "median", c.prettyKey( c.currKey() ) );
+            BSONObj median = c.prettyKey( c.currKey() );
+            result.append( "median", median );
+
+            int x = median.woCompare( min , BSONObj() , false );
+            int y = median.woCompare( max , BSONObj() , false );
+            if ( x == 0 || y == 0 ){
+                // its on an edge, ok
+            }
+            else if ( x < 0 && y < 0 ){
+                log( LL_ERROR ) << "median error (1) min: " << min << " max: " << max << " median: " << median << endl;
+                errmsg = "median error 1";
+                return false;
+            }
+            else if ( x > 0 && y > 0 ){
+                log( LL_ERROR ) << "median error (2) min: " << min << " max: " << max << " median: " << median << endl;
+                errmsg = "median error 2";
+                return false;
+            }
+
             return true;
         }
     } cmdMedianKey;