added 'verify writes' switch to track down data corruption

Parms.cpp

@@ -11297,7 +11297,7 @@ void Parms::init ( ) {
 	m->m_type  = TYPE_BOOL;
 	m->m_def   = "0";
 	m->m_group = 0;
-	m->m_flags = PF_HIDDEN | PF_NOSAVE;
+	m->m_flags = 0;//PF_HIDDEN | PF_NOSAVE;
 	m->m_page  = PAGE_MASTER;
 	m->m_obj   = OBJ_CONF;
 	m++;

RdbDump.cpp

@@ -405,12 +405,15 @@ bool RdbDump::dumpTree ( bool recall ) {
 		m_totalNegDumped += m_numNegRecs;
 		// . check the list we got from the tree for problems
 		// . ensures keys are ordered from lowest to highest as well
-#ifdef GBSANITYCHECK
-		log("dump: verifying list before dumping");
-		m_list->checkList_r ( false , // removeNegRecs?
-				      false , // sleep on problem?
-				      m_rdb->m_rdbId );
-#endif
+		//#ifdef GBSANITYCHECK
+		if ( g_conf.m_verifyWrites ) {
+			char *s = "none";
+			if ( m_rdb ) s = getDbnameFromId(m_rdb->m_rdbId);
+			log("dump: verifying list before dumping (rdb=%s)",s);
+			m_list->checkList_r ( false , // removeNegRecs?
+					      false , // sleep on problem?
+					      m_rdb->m_rdbId );
+		}
 		// if list is empty, we're done!
 		if ( status && m_list->isEmpty() ) {
 			// consider that a rollover?
@@ -486,15 +489,15 @@ bool RdbDump::dumpList ( RdbList *list , int32_t niceness , bool recall ) {
 	if ( m_list->isEmpty() ) return true;
 	// we're now in dump mode again 
 	m_isDumping = true;
-#ifdef GBSANITYCHECK
+	//#ifdef GBSANITYCHECK
 	// don't check list if we're dumping an unordered list from tree!
-	if ( m_orderedDump ) {
+	if ( g_conf.m_verifyWrites && m_orderedDump ) {
 		m_list->checkList_r ( false /*removedNegRecs?*/ );
 		// print list stats
 		log("dump: sk=%s ",KEYSTR(m_list->m_startKey,m_ks));
 		log("dump: ek=%s ",KEYSTR(m_list->m_endKey,m_ks));
 	}
-#endif
+	//#endif
 
 	// before calling RdbMap::addList(), always reset list ptr
 	// since we no longer call this in RdbMap::addList() so we don't
@@ -525,8 +528,10 @@ bool RdbDump::dumpList ( RdbList *list , int32_t niceness , bool recall ) {
 		}
 	}
 
-	if ( m_ks==18 ) {
-		m_list->checkList_r(false,false,RDB_POSDB);
+	if ( g_conf.m_verifyWrites ) {
+		char rdbId = 0;
+		if ( m_rdb ) rdbId = m_rdb->m_rdbId;
+		m_list->checkList_r(false,false,rdbId);//RDB_POSDB);
 		m_list->resetListPtr();
 	}
 

RdbList.cpp

@@ -693,9 +693,9 @@ bool RdbList::checkList_r ( bool removeNegRecs , bool sleepOnProblem ,
 		return false;
 	}
 
-	if ( m_useHalfKeys && m_ks == 12 ) // m_ks != 18 && m_ks != 24 ) 
-		return checkIndexList_r ( removeNegRecs  , 
-					  sleepOnProblem );
+	// if ( m_useHalfKeys && m_ks == 12 ) // m_ks != 18 && m_ks != 24 ) 
+	// 	return checkIndexList_r ( removeNegRecs  , 
+	// 				  sleepOnProblem );
 
 	//log("m_list=%"INT32"",(int32_t)m_list);
 	//key_t oldk;
@@ -721,6 +721,10 @@ bool RdbList::checkList_r ( bool removeNegRecs , bool sleepOnProblem ,
 	if ( KEYCMP(acceptable,KEYMIN(),m_ks)==0 )
 		KEYSET ( acceptable , m_endKey , m_ks );
 	char k[MAX_KEY_BYTES];
+
+	static int32_t th = 0;
+	if ( ! th ) th = hash64Lower_a ( "roottitles" , 10 );
+
 	while ( ! isExhausted() ) {
 		//key_t k = getCurrentKey();
 		getCurrentKey( k );
@@ -734,6 +738,43 @@ bool RdbList::checkList_r ( bool removeNegRecs , bool sleepOnProblem ,
 			      *(int32_t *)data > 100000000 ) ) {
 				char *xx = NULL; *xx = 0; }
 		}		
+		// tagrec?
+		if ( rdbId == RDB_TAGDB && ! KEYNEG(k) ) {
+			//TagRec *gr = (TagRec *)getCurrentRec();
+			//Tag *tag = gr->getFirstTag   ( );
+			//for ( ; tag ; tag = gr->getNextTag ( tag ) ) {
+			Tag *tag = (Tag *)getCurrentRec();
+			if ( tag->m_type == th ) {
+				char *tdata = tag->getTagData();
+				int32_t tsize = tag->getTagDataSize();
+				// core if tag val is not \0 terminated
+				if ( tsize > 0 && tdata[tsize-1]!='\0' ) {
+					log("db: bad root title tag");
+					char *xx=NULL;*xx=0; }
+			}
+		}
+		if ( rdbId == RDB_SPIDERDB && ! KEYNEG(k) &&
+		     getCurrentDataSize() > 0 ) {
+			//char *data = getCurrentData();
+			char *rec = getCurrentRec();
+			// bad url in spider request?
+			if ( g_spiderdb.isSpiderRequest ( (key128_t *)rec ) ){
+				SpiderRequest *sr = (SpiderRequest *)rec;
+				if ( strncmp(sr->m_url,"http",4) != 0 ) {
+					log("db: spider req url");
+					char *xx=NULL;*xx=0;
+				}
+			}
+		}
+		// title bad uncompress size?
+		if ( rdbId == RDB_TITLEDB && ! KEYNEG(k) ) {
+			char *rec = getCurrentRec();
+			int32_t usize = *(int32_t *)(rec+12+4);
+			if ( usize <= 0 ) {
+				log("db: bad titlerec uncompress size");
+				char *xx=NULL;*xx=0; 
+			}
+		}
 		// debug msg
 		// pause if it's google
 		//if ((((k.n0)  >> 1) & 0x0000003fffffffffLL)  == 70166155664) 

RdbTree.cpp

@@ -2495,6 +2495,20 @@ bool RdbTree::fastSave_r() {
 		return log("db: Could not open %s for writing: %s.",
 			   s,mstrerror(errno));
 	}
+
+ redo:
+	// verify the tree
+	if ( g_conf.m_verifyWrites ) {
+		log("db: verify writes is enabled, checking tree before "
+		    "saving.");
+		if ( ! checkTree( false , true ) ) {
+			log("db: fixing tree and re-checking");
+			fixTree ( );
+			goto redo;
+		}
+	}
+
+
 	// clear our own errno
 	errno = 0;
 	// . save the header