pad objects when allocated, when appropriate, to make updates faster.

3 files changed, 41 insertions, 6 deletions

diff --git a/db/namespace.h b/db/namespace.h
index 6caefc633f3..e8e056ed958 100644
--- a/db/namespace.h
+++ b/db/namespace.h
@@ -76,11 +76,14 @@ extern int bucketSizes[];
 class NamespaceDetails {
 public:
 	NamespaceDetails() { 
+		/* be sure to initialize new fields here, otherwise they will contain garbage by default.
+		*/
 		datasize = nrecords = 0;
 		lastExtentSize = 0;
 		nIndexes = 0;
 		capped = 0;
 		max = 0x7fffffff;
+		paddingFactor = 1.0;
 		memset(reserved, 0, sizeof(reserved));
 	} 
 	DiskLoc firstExtent;
@@ -93,7 +96,19 @@ public:
 	IndexDetails indexes[MaxIndexes];
 	int capped;
 	int max; // max # of objects for a capped table.
-	char reserved[256-16-4-4-8*MaxIndexes-8-8];
+	double paddingFactor; // 1.0 = no padding.
+	char reserved[256-16-4-4-8*MaxIndexes-8-8-8];
+
+	void paddingFits() { 
+		double x = paddingFactor - 0.01;
+		if( x >= 1.0 )
+			paddingFactor = x;
+	}
+	void paddingTooSmall() { 
+		double x = paddingFactor + 0.6;
+		if( x <= 2.0 )
+			paddingFactor = x;
+	}
 
 	//returns offset in indexes[]
 	int findIndexByName(const char *name) { 
@@ -112,8 +127,12 @@ public:
 		return Buckets-1;
 	}
 
+	/* allocate a new record.  lenToAlloc includes headers. */
 	DiskLoc alloc(const char *ns, int lenToAlloc, DiskLoc& extentLoc);
+
+	/* add a given record to the deleted chains for this NS */
 	void addDeletedRec(DeletedRecord *d, DiskLoc dloc);
+
 	void dumpDeleted(set<DiskLoc> *extents = 0);
 private:
 	DiskLoc __stdAlloc(int len);
diff --git a/db/pdfile.cpp b/db/pdfile.cpp
index 2bc7c529bdf..c39191d42a7 100644
--- a/db/pdfile.cpp
+++ b/db/pdfile.cpp
@@ -79,7 +79,8 @@ void NamespaceDetails::addDeletedRec(DeletedRecord *d, DiskLoc dloc) {
 	d->nextDeleted = oldHead;
 }
 
-/* lenToAlloc is WITH header 
+/* 
+   lenToAlloc is WITH header 
 */
 DiskLoc NamespaceDetails::alloc(const char *ns, int lenToAlloc, DiskLoc& extentLoc) {
 	lenToAlloc = (lenToAlloc + 3) & 0xfffffffc;
@@ -259,10 +260,13 @@ void NamespaceDetails::compact() {
 	}
 }
 
+/* alloc with capped table handling. */
 DiskLoc NamespaceDetails::_alloc(const char *ns, int len) {
 	if( !capped )
 		return __stdAlloc(len);
 
+	// capped. 
+
 	assert( len < 400000000 );
 	int passes = 0;
 	DiskLoc loc;
@@ -749,13 +753,15 @@ void DataFileMgr::update(
 {
 	NamespaceDetails *d = nsdetails(ns);
 
-	if( toupdate->netLength() < len ) {
+	if(  toupdate->netLength() < len ) {
+		// doesn't fit.  must reallocate.
+
 		if( d && d->capped ) { 
 			ss << " failing a growing update on a capped ns " << ns << endl;
 			return;
 		}
 
-		// doesn't fit.
+		d->paddingTooSmall();
 		if( client->profile )
 			ss << " moved ";
 		deleteRecord(ns, toupdate, dl);
@@ -763,6 +769,8 @@ void DataFileMgr::update(
 		return;
 	}
 
+	d->paddingFits();
+
 	/* has any index keys changed? */
 	{
 		NamespaceDetails *d = nsdetails(ns);
@@ -893,6 +901,7 @@ DiskLoc DataFileMgr::insert(const char *ns, const void *buf, int len, bool god)
 		client->newestFile()->newExtent(ns, initialExtentSize(len));
 		d = nsdetails(ns);
 	}
+	d->paddingFits();
 
 	NamespaceDetails *tableToIndex = 0;
 
@@ -927,11 +936,18 @@ DiskLoc DataFileMgr::insert(const char *ns, const void *buf, int len, bool god)
 
 	DiskLoc extentLoc;
 	int lenWHdr = len + Record::HeaderSize;
+	lenWHdr = (int) (lenWHdr * d->paddingFactor);
+	if( lenWHdr == 0 ) {
+		// old datafiles, backward compatible here.
+		assert( d->paddingFactor == 0 );
+		d->paddingFactor = 1.0;
+		lenWHdr = len + Record::HeaderSize;
+	}
 	DiskLoc loc = d->alloc(ns, lenWHdr, extentLoc);
 	if( loc.isNull() ) {
 		// out of space
 		if( d->capped == 0 ) { // size capped doesn't grow
-			cout << "allocating new extent for " << ns << endl;
+			cout << "allocating new extent for " << ns << " padding:" << d->paddingFactor << endl;
 			client->newestFile()->newExtent(ns, followupExtentSize(len, d->lastExtentSize));
 			loc = d->alloc(ns, lenWHdr, extentLoc);
 		}
diff --git a/db/query.cpp b/db/query.cpp
index cef5dbf50a9..3f76624dfb3 100644
--- a/db/query.cpp
+++ b/db/query.cpp
@@ -342,7 +342,7 @@ string validateNS(const char *ns, NamespaceDetails *d) {
 
 	ss << "  firstExtent:" << d->firstExtent.toString() << " lastExtent:" << d->lastExtent.toString() << '\n';
 	ss << "  datasize?:" << d->datasize << " nrecords?:" << d->nrecords << " lastExtentSize:" << d->lastExtentSize << '\n';
-
+	ss << "  padding:" << d->paddingFactor << '\n';
 	try { 
 
 		{