From: Date: March 20 2007 4:16pm
Subject: bk commit into 5.0 tree (jonas:1.2298) BUG#27283
List-Archive: http://lists.mysql.com/commits/22367
X-Bug: 27283
Message-Id: <20070320151629.542C383677@perch.ndb.mysql.com>

Below is the list of changes that have just been committed into a local
5.0 repository of jonas. When jonas does a push these changes will
be propagated to the main repository and, within 24 hours after the
push, to the public repository.
For information on how to access the public repository
see http://dev.mysql.com/doc/mysql/en/installing-source-tree.html

ChangeSet@stripped, 2007-03-20 16:16:25+01:00, jonas@stripped +4 -0
  ndb - bug#27283 (wl2325-5.0)
      Handle race condtition between MASTER_GCPCONF and execGCP_NODEFINISH

  ndb/src/kernel/blocks/ERROR_codes.txt@stripped, 2007-03-20 16:16:24+01:00, jonas@stripped +1 -1
    new error codes

  ndb/src/kernel/blocks/dbdih/DbdihMain.cpp@stripped, 2007-03-20 16:16:24+01:00, jonas@stripped +29 -3
    Handle race condtition between MASTER_GCPCONF and execGCP_NODEFINISH

  ndb/test/ndbapi/testNodeRestart.cpp@stripped, 2007-03-20 16:16:24+01:00, jonas@stripped +45 -0
    testcase

  ndb/test/run-test/daily-basic-tests.txt@stripped, 2007-03-20 16:16:24+01:00, jonas@stripped +4 -0
    testcase

# This is a BitKeeper patch.  What follows are the unified diffs for the
# set of deltas contained in the patch.  The rest of the patch, the part
# that BitKeeper cares about, is below these diffs.
# User:	jonas
# Host:	perch.ndb.mysql.com
# Root:	/home/jonas/src/50-work

--- 1.50/ndb/test/run-test/daily-basic-tests.txt	2007-03-20 16:16:29 +01:00
+++ 1.51/ndb/test/run-test/daily-basic-tests.txt	2007-03-20 16:16:29 +01:00
@@ -429,6 +429,10 @@
 cmd: testNodeRestart
 args: -n Bug27003 T1
 
+max-time: 1000
+cmd: testNodeRestart
+args: -n Bug27283 T1
+
 max-time: 500
 cmd: testNodeRestart
 args: -n Bug15587 T1

--- 1.27/ndb/src/kernel/blocks/ERROR_codes.txt	2007-03-20 16:16:29 +01:00
+++ 1.28/ndb/src/kernel/blocks/ERROR_codes.txt	2007-03-20 16:16:29 +01:00
@@ -5,7 +5,7 @@
 Next DBTUP 4014
 Next DBLQH 5043
 Next DBDICT 6007
-Next DBDIH 7181
+Next DBDIH 7183
 Next DBTC 8039
 Next CMVMI 9000
 Next BACKUP 10022

--- 1.68/ndb/src/kernel/blocks/dbdih/DbdihMain.cpp	2007-03-20 16:16:29 +01:00
+++ 1.69/ndb/src/kernel/blocks/dbdih/DbdihMain.cpp	2007-03-20 16:16:29 +01:00
@@ -4811,6 +4811,15 @@
   } else {
     ndbrequire(failedNodePtr.p->nodeStatus == NodeRecord::DYING);
   }//if
+
+  if (ERROR_INSERTED(7181))
+  {
+    ndbout_c("execGCP_TCFINISHED in MASTER_GCPREQ");
+    CLEAR_ERROR_INSERT_VALUE;
+    signal->theData[1] = coldgcp;
+    execGCP_TCFINISHED(signal);
+  }
+  
   MasterGCPConf::State gcpState;
   switch (cgcpParticipantState) {
   case GCP_PARTICIPANT_READY:
@@ -4877,6 +4886,14 @@
     masterGCPConf->lcpActive[i] = SYSFILE->lcpActive[i];
   sendSignal(newMasterBlockref, GSN_MASTER_GCPCONF, signal, 
              MasterGCPConf::SignalLength, JBB);
+
+  if (ERROR_INSERTED(7182))
+  {
+    ndbout_c("execGCP_TCFINISHED in MASTER_GCPREQ");
+    CLEAR_ERROR_INSERT_VALUE;
+    signal->theData[1] = coldgcp;
+    execGCP_TCFINISHED(signal);
+  }
 }//Dbdih::execMASTER_GCPREQ()
 
 void Dbdih::execMASTER_GCPCONF(Signal* signal) 
@@ -7542,10 +7559,10 @@
   } else if (cmasterState == MASTER_TAKE_OVER_GCP) {
     jam();
     //-------------------------------------------------------------
-    // We are currently taking over as master. We will delay the
-    // signal until we have completed the take over gcp handling.
+    // We are currently taking over as master. Ignore
+    // signal in this case since we will discover it in reception of 
+    // MASTER_GCPCONF.
     //-------------------------------------------------------------
-    sendSignalWithDelay(reference(), GSN_GCP_NODEFINISH, signal, 20, 3);
     return;
   } else {
     ndbrequire(cmasterState == MASTER_ACTIVE);
@@ -7691,6 +7708,15 @@
   CRASH_INSERTION(7007);
   Uint32 gci = signal->theData[1];
   ndbrequire(gci == coldgcp);
+
+  if (ERROR_INSERTED(7181) || ERROR_INSERTED(7182))
+  {
+    ndbout_c("killing %d", refToNode(cmasterdihref));
+    signal->theData[0] = 9999;
+    sendSignal(numberToRef(CMVMI, refToNode(cmasterdihref)),
+	       GSN_NDB_TAMPER, signal, 1, JBB);
+    return;
+  }
 
   cgcpParticipantState = GCP_PARTICIPANT_TC_FINISHED;
   signal->theData[0] = cownNodeId;

--- 1.29/ndb/test/ndbapi/testNodeRestart.cpp	2007-03-20 16:16:29 +01:00
+++ 1.30/ndb/test/ndbapi/testNodeRestart.cpp	2007-03-20 16:16:29 +01:00
@@ -1178,6 +1178,48 @@
 }
 
 
+int
+runBug27283(NDBT_Context* ctx, NDBT_Step* step)
+{
+  int result = NDBT_OK;
+  int loops = ctx->getNumLoops();
+  int records = ctx->getNumRecords();
+  NdbRestarter res;
+
+  if (res.getNumDbNodes() < 2)
+  {
+    return NDBT_OK;
+  }
+
+  static const int errnos[] = { 7181, 7182, 0 };
+  
+  Uint32 pos = 0;
+  for (Uint32 i = 0; i<loops; i++)
+  {
+    while (errnos[pos] != 0)
+    {
+      int master = res.getMasterNodeId();
+      int next = res.getNextMasterNodeId(master);
+      int next2 = res.getNextMasterNodeId(next);
+      
+      int node = (i & 1) ? next : next2;
+      ndbout_c("Tesing err: %d", errnos[pos]);
+      if (res.insertErrorInNode(next, errnos[pos]))
+	return NDBT_FAILED;
+
+      NdbSleep_SecSleep(3);
+      
+      if (res.waitClusterStarted())
+	return NDBT_FAILED;
+      
+      pos++;
+    }
+    pos = 0;
+  }
+  
+  return NDBT_OK;
+}
+
 NDBT_TESTSUITE(testNodeRestart);
 TESTCASE("NoLoad", 
 	 "Test that one node at a time can be stopped and then restarted "\
@@ -1507,6 +1549,9 @@
 }
 TESTCASE("Bug27003", ""){
   INITIALIZER(runBug27003);
+}
+TESTCASE("Bug27283", ""){
+  INITIALIZER(runBug27283);
 }
 NDBT_TESTSUITE_END(testNodeRestart);