Merge perch.ndb.mysql.com:/home/jonas/src/mysql-4.1 (27388715) · Commits · Software / OSDI20 Artifacts / mariadb

ndb/include/kernel/signaldata/DumpStateOrd.hpp

+2 −0

Original line number	Diff line number	Diff line
		@@ -78,6 +78,8 @@ public:
		LqhDumpAllScanRec = 2301,
		LqhDumpAllActiveScanRec = 2302,
		LqhDumpLcpState = 2303,
		LqhErrorInsert5042 = 2315,

		AccDumpOneScanRec = 2400,
		AccDumpAllScanRec = 2401,
		AccDumpAllActiveScanRec = 2402,

ndb/include/ndb_version.h.in

+3 −0

Original line number	Diff line number	Diff line
		@@ -57,5 +57,8 @@ char ndb_version_string_buf[NDB_VERSION_STRING_BUF_SZ];
		*/
		/#define NDB_VERSION_ID 0/

		#define NDBD_INCL_NODECONF_VERSION_4 MAKE_VERSION(4,1,17)
		#define NDBD_INCL_NODECONF_VERSION_5 MAKE_VERSION(5,0,18)

		#endif

ndb/src/kernel/blocks/ERROR_codes.txt

+9 −0

Original line number	Diff line number	Diff line
		@@ -61,6 +61,8 @@ Insert system error in GCP participant when receiving GCP_SAVEREQ.
		5007:
		Delay GCP_SAVEREQ by 10 secs

		7165: Delay INCL_NODE_REQ in starting node yeilding error in GCP_PREPARE

		ERROR CODES FOR TESTING NODE FAILURE, LOCAL CHECKPOINT HANDLING:
		-----------------------------------------------------------------

		@@ -155,11 +157,15 @@ Insert node failure handling when receiving COMPLETEREQ.
		5006:
		Insert node failure handling when receiving ABORTREQ.

		5042:
		As 5002, but with specified table (see DumpStateOrd)

		These error code can be combined with error codes for testing time-out
		handling in DBTC to ensure that node failures are also well handled in
		time-out handling. They can also be used to test multiple node failure
		handling.


		ERROR CODES FOR TESTING TIME-OUT HANDLING IN DBLQH
		-------------------------------------------------
		5011:
		@@ -196,6 +202,9 @@ Delay execution of ABORTREQ signal 2 seconds to generate time-out.
		8048: Make TC not choose own node for simple/dirty read
		5041: Crash is receiving simple read from other TC on different node

		5100,5101: Drop ABORT req in primary replica
		Crash on "next" ABORT

		ERROR CODES FOR TESTING TIME-OUT HANDLING IN DBTC
		-------------------------------------------------
		8040:

ndb/src/kernel/blocks/dbdih/DbdihMain.cpp

+47 −5

Original line number	Diff line number	Diff line
		@@ -215,7 +215,7 @@ void Dbdih::sendINCL_NODEREQ(Signal* signal, Uint32 nodeId)
		signal->theData[2] = c_nodeStartMaster.failNr;
		signal->theData[3] = 0;
		signal->theData[4] = currentgcp;
		sendSignal(nodeDihRef, GSN_INCL_NODEREQ, signal, 5, JBB);
		sendSignal(nodeDihRef, GSN_INCL_NODEREQ, signal, 5, JBA);
		}//Dbdih::sendINCL_NODEREQ()

		void Dbdih::sendMASTER_GCPREQ(Signal* signal, Uint32 nodeId)
		@@ -1857,6 +1857,14 @@ void Dbdih::gcpBlockedLab(Signal* signal)
		// global checkpoint id and the correct state. We do not wait for any reply
		// since the starting node will not send any.
		/-------------------------------------------------------------------------/
		Uint32 startVersion = getNodeInfo(c_nodeStartMaster.startNode).m_version;

		if ((getMajor(startVersion) == 4 && startVersion >= NDBD_INCL_NODECONF_VERSION_4) \|\|
		(getMajor(startVersion) == 5 && startVersion >= NDBD_INCL_NODECONF_VERSION_5))
		{
		c_INCL_NODEREQ_Counter.setWaitingFor(c_nodeStartMaster.startNode);
		}

		sendINCL_NODEREQ(signal, c_nodeStartMaster.startNode);
		}//Dbdih::gcpBlockedLab()

		@@ -2059,6 +2067,13 @@ void Dbdih::execINCL_NODEREQ(Signal* signal)
		jamEntry();
		Uint32 retRef = signal->theData[0];
		Uint32 nodeId = signal->theData[1];
		if (nodeId == getOwnNodeId() && ERROR_INSERTED(7165))
		{
		CLEAR_ERROR_INSERT_VALUE;
		sendSignalWithDelay(reference(), GSN_INCL_NODEREQ, signal, 5000, signal->getLength());
		return;
		}

		Uint32 tnodeStartFailNr = signal->theData[2];
		currentgcp = signal->theData[4];
		CRASH_INSERTION(7127);
		@@ -2086,6 +2101,15 @@ void Dbdih::execINCL_NODEREQ(Signal* signal)
		// id's and the lcp status.
		/-----------------------------------------------------------------------/
		CRASH_INSERTION(7171);
		Uint32 masterVersion = getNodeInfo(refToNode(cmasterdihref)).m_version;

		if ((NDB_VERSION_MAJOR == 4 && masterVersion >= NDBD_INCL_NODECONF_VERSION_4) \|\|
		(NDB_VERSION_MAJOR == 5 && masterVersion >= NDBD_INCL_NODECONF_VERSION_5))
		{
		signal->theData[0] = getOwnNodeId();
		signal->theData[1] = getOwnNodeId();
		sendSignal(cmasterdihref, GSN_INCL_NODECONF, signal, 2, JBB);
		}
		return;
		}//if
		if (getNodeStatus(nodeId) != NodeRecord::STARTING) {
		@@ -3737,8 +3761,16 @@ void Dbdih::execNODE_FAILREP(Signal* signal)
		/------------------------------------------------------------------------/
		// Verify that a starting node has also crashed. Reset the node start record.
		/-------------------------------------------------------------------------/
		if (c_nodeStartMaster.startNode != RNIL) {
		ndbrequire(getNodeStatus(c_nodeStartMaster.startNode)!= NodeRecord::ALIVE);
		if (false && c_nodeStartMaster.startNode != RNIL && getNodeStatus(c_nodeStartMaster.startNode) == NodeRecord::ALIVE)
		{
		BlockReference cntrRef = calcNdbCntrBlockRef(c_nodeStartMaster.startNode);
		SystemError * const sysErr = (SystemError*)&signal->theData[0];
		sysErr->errorCode = SystemError::StartInProgressError;
		sysErr->errorRef = reference();
		sysErr->data1= 0;
		sysErr->data2= __LINE__;
		sendSignal(cntrRef, GSN_SYSTEM_ERROR, signal, SystemError::SignalLength, JBA);
		nodeResetStart();
		}//if

		/--------------------------------------------------/
		@@ -5187,15 +5219,16 @@ void Dbdih::removeNodeFromTable(Signal* signal,
		/**
		* For each of replica record
		*/
		Uint32 replicaNo = 0;
		bool found = false;
		ReplicaRecordPtr replicaPtr;
		for(replicaPtr.i = fragPtr.p->storedReplicas; replicaPtr.i != RNIL;
		replicaPtr.i = replicaPtr.p->nextReplica, replicaNo++) {
		replicaPtr.i = replicaPtr.p->nextReplica) {
		jam();

		ptrCheckGuard(replicaPtr, creplicaFileSize, replicaRecord);
		if(replicaPtr.p->procNode == nodeId){
		jam();
		found = true;
		noOfRemovedReplicas++;
		removeNodeFromStored(nodeId, fragPtr, replicaPtr);
		if(replicaPtr.p->lcpOngoingFlag){
		@@ -5211,6 +5244,15 @@ void Dbdih::removeNodeFromTable(Signal* signal,
		}
		}
		}
		if (!found)
		{
		jam();
		/**
		* Run updateNodeInfo to remove any dead nodes from list of activeNodes
		* see bug#15587
		*/
		updateNodeInfo(fragPtr);
		}
		noOfRemainingLcpReplicas += fragPtr.p->noLcpReplicas;
		}

ndb/src/kernel/blocks/dblqh/Dblqh.hpp

+1 −0

Original line number	Diff line number	Diff line
		@@ -2881,6 +2881,7 @@ private:
		UintR ctransidHash[1024];

		Uint32 c_diskless;
		Uint32 c_error_insert_table_id;

		public:
		/**