Below is the list of changes that have just been committed into a local
5.0 repository of tomas. When tomas 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, 2007-05-28 15:59:02+02:00, tomas@stripped +5 -0
bug#28717, make sure only master updates activeStatus
so that othernodes dont get confused after having recevied status from master
and then tries to update it self
ndb/src/kernel/blocks/ERROR_codes.txt@stripped, 2007-05-28 15:59:00+02:00, tomas@stripped +6 -1
error 1001, delay node_failrep
ndb/src/kernel/blocks/dbdih/DbdihMain.cpp@stripped, 2007-05-28 15:59:00+02:00, tomas@stripped +7 -1
bug#28717, make sure only master updates activeStatus
so that othernodes dont get confused after having recevied status from master
and then tries to update it self
ndb/src/kernel/blocks/ndbcntr/NdbcntrMain.cpp@stripped, 2007-05-28 15:59:00+02:00, tomas@stripped +7 -0
error 1001, delay node_failrep
ndb/test/ndbapi/testNodeRestart.cpp@stripped, 2007-05-28 15:59:01+02:00, tomas@stripped +81 -0
testcase
ndb/test/run-test/daily-basic-tests.txt@stripped, 2007-05-28 15:59:01+02:00, tomas@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: tomas
# Host: whalegate.ndb.mysql.com
# Root: /home/tomas/mysql-5.0-telco-gca
--- 1.47/ndb/test/run-test/daily-basic-tests.txt 2007-04-25 09:23:26 +02:00
+++ 1.48/ndb/test/run-test/daily-basic-tests.txt 2007-05-28 15:59:01 +02:00
@@ -492,6 +492,10 @@
cmd: testDict
args: -n CreateAndDrop
+max-time: 1000
+cmd: testNodeRestart
+args: -n Bug28717 T1
+
max-time: 1500
cmd: testDict
args: -n CreateAndDropAtRandom -l 200 T1
--- 1.25/ndb/src/kernel/blocks/ERROR_codes.txt 2007-01-03 06:17:32 +01:00
+++ 1.26/ndb/src/kernel/blocks/ERROR_codes.txt 2007-05-28 15:59:00 +02:00
@@ -1,5 +1,5 @@
Next QMGR 1
-Next NDBCNTR 1000
+Next NDBCNTR 1002
Next NDBFS 2000
Next DBACC 3002
Next DBTUP 4014
@@ -487,3 +487,8 @@
6003 Crash in participant @ CreateTabReq::Prepare
6004 Crash in participant @ CreateTabReq::Commit
6005 Crash in participant @ CreateTabReq::CreateDrop
+
+Ndbcntr:
+--------
+
+1001: Delay sending NODE_FAILREP (to own node), until error is cleared
--- 1.67/ndb/src/kernel/blocks/dbdih/DbdihMain.cpp 2007-03-07 02:11:07 +01:00
+++ 1.68/ndb/src/kernel/blocks/dbdih/DbdihMain.cpp 2007-05-28 15:59:00 +02:00
@@ -4448,12 +4448,18 @@
jam();
const Uint32 nodeId = failedNodePtr.i;
- if (c_lcpState.m_participatingLQH.get(failedNodePtr.i)){
+ if (isMaster() && c_lcpState.m_participatingLQH.get(failedNodePtr.i))
+ {
/*----------------------------------------------------*/
/* THE NODE WAS INVOLVED IN A LOCAL CHECKPOINT. WE */
/* MUST UPDATE THE ACTIVE STATUS TO INDICATE THAT */
/* THE NODE HAVE MISSED A LOCAL CHECKPOINT. */
/*----------------------------------------------------*/
+
+ /**
+ * Bug#28717, Only master should do this, as this status is copied
+ * to other nodes
+ */
switch (failedNodePtr.p->activeStatus) {
case Sysfile::NS_Active:
jam();
--- 1.35/ndb/src/kernel/blocks/ndbcntr/NdbcntrMain.cpp 2006-12-23 20:04:16 +01:00
+++ 1.36/ndb/src/kernel/blocks/ndbcntr/NdbcntrMain.cpp 2007-05-28 15:59:00 +02:00
@@ -1375,6 +1375,13 @@
{
jamEntry();
+ if (ERROR_INSERTED(1001))
+ {
+ sendSignalWithDelay(reference(), GSN_NODE_FAILREP, signal, 100,
+ signal->getLength());
+ return;
+ }
+
const NodeFailRep * nodeFail = (NodeFailRep *)&signal->theData[0];
NdbNodeBitmask allFailed;
allFailed.assign(NdbNodeBitmask::Size, nodeFail->theNodes);
--- 1.26/ndb/test/ndbapi/testNodeRestart.cpp 2007-01-11 21:12:09 +01:00
+++ 1.27/ndb/test/ndbapi/testNodeRestart.cpp 2007-05-28 15:59:01 +02:00
@@ -1045,6 +1045,84 @@
}
+int
+runBug28717(NDBT_Context* ctx, NDBT_Step* step)
+{
+ int result = NDBT_OK;
+ int loops = ctx->getNumLoops();
+ int records = ctx->getNumRecords();
+ Ndb* pNdb = GETNDB(step);
+ NdbRestarter res;
+
+ if (res.getNumDbNodes() < 4)
+ {
+ return NDBT_OK;
+ }
+
+ int master = res.getMasterNodeId();
+ int node0 = res.getRandomNodeOtherNodeGroup(master, rand());
+ int node1 = res.getRandomNodeSameNodeGroup(node0, rand());
+
+ ndbout_c("master: %d node0: %d node1: %d", master, node0, node1);
+
+ if (res.restartOneDbNode(node0, false, true, true))
+ {
+ return NDBT_FAILED;
+ }
+
+ {
+ int filter[] = { 15, NDB_MGM_EVENT_CATEGORY_CHECKPOINT, 0 };
+ NdbLogEventHandle handle =
+ ndb_mgm_create_logevent_handle(res.handle, filter);
+
+
+ int dump[] = { DumpStateOrd::DihStartLcpImmediately };
+ struct ndb_logevent event;
+
+ for (Uint32 i = 0; i<3; i++)
+ {
+ res.dumpStateOneNode(master, dump, 1);
+ while(ndb_logevent_get_next(handle, &event, 0) >= 0 &&
+ event.type != NDB_LE_LocalCheckpointStarted);
+ while(ndb_logevent_get_next(handle, &event, 0) >= 0 &&
+ event.type != NDB_LE_LocalCheckpointCompleted);
+ }
+ }
+
+ if (res.waitNodesNoStart(&node0, 1))
+ return NDBT_FAILED;
+
+ int val2[] = { DumpStateOrd::CmvmiSetRestartOnErrorInsert, 1 };
+
+ if (res.dumpStateOneNode(node0, val2, 2))
+ return NDBT_FAILED;
+
+ if (res.insertErrorInNode(node0, 5010))
+ return NDBT_FAILED;
+
+ if (res.insertErrorInNode(node1, 1001))
+ return NDBT_FAILED;
+
+ if (res.startNodes(&node0, 1))
+ return NDBT_FAILED;
+
+ NdbSleep_SecSleep(3);
+
+ if (res.insertErrorInNode(node1, 0))
+ return NDBT_FAILED;
+
+ if (res.waitNodesNoStart(&node0, 1))
+ return NDBT_FAILED;
+
+ if (res.startNodes(&node0, 1))
+ return NDBT_FAILED;
+
+ if (res.waitClusterStarted())
+ return NDBT_FAILED;
+
+ return NDBT_OK;
+}
+
NDBT_TESTSUITE(testNodeRestart);
TESTCASE("NoLoad",
"Test that one node at a time can be stopped and then restarted "\
@@ -1365,6 +1443,9 @@
}
TESTCASE("Bug25554", ""){
INITIALIZER(runBug25554);
+}
+TESTCASE("Bug28717", ""){
+ INITIALIZER(runBug28717);
}
NDBT_TESTSUITE_END(testNodeRestart);
| Thread |
|---|
| • bk commit into 5.0 tree (tomas:1.2404) BUG#28717 | tomas | 28 May |
