Add shutdown options (so far only SIGKILL), so that there won't be any strace mysqld threads
ndb/src/cw/cpcd/APIService.cpp:
Add shutdown options
(so far only SIGKILL)
ndb/src/cw/cpcd/CPCD.hpp:
Add shutdown options
(so far only SIGKILL)
ndb/src/cw/cpcd/Process.cpp:
Add shutdown options
(so far only SIGKILL)
ndb/test/include/CpcClient.hpp:
Add shutdown options
(so far only SIGKILL)
ndb/test/run-test/main.cpp:
Add shutdown options
(so far only SIGKILL)
ndb/test/src/CpcClient.cpp:
Add shutdown options
(so far only SIGKILL)
put config files in bk
ndb/test/run-test/make-config.sh:
Corrent directory
ndb/test/run-test/ndb-autotest.sh:
Check for 5.1 disrectory structure
Add config files to distribution
Remove unpacking for ndb-autotest.sh
use newly created script for faster handling of several configurations
ndb/test/run-test/make-config.sh:
put cluster config in same config as distribution config
ndb/test/run-test/ndb-autotest.sh:
1) install mysql_db
2) use make-config script
increase no of rows so that the node failure will "guaranteed" produce an aborted transaction
ndb/test/ndbapi/testNodeRestart.cpp:
increase no of rows so that the node failure will "guaranteed" produce an aborted transaction
Cluster failure with non started nodes can result in timedout transactions
ndb/src/mgmapi/mgmapi.cpp:
Increase timeout for restarts
ndb/src/ndbapi/ClusterMgr.cpp:
Report NFCOMPLETEREP if no alive node exists
(instead of no connected node exists)
ndb/src/ndbapi/ClusterMgr.hpp:
Report NFCOMPLETEREP if no alive node exists
(instead of no connected node exists)
ndb/src/kernel/blocks/dbtc/Dbtc.hpp:
remove unneccessary TcSeizedIndexOperation object
ndb/src/kernel/blocks/dbtc/DbtcInit.cpp:
remove unneccessary TcSeizedIndexOperation object
ndb/src/kernel/blocks/dbtc/DbtcMain.cpp:
1) Set transid on starting TCINDXREQ
2) New error code for out of index operation records
3) Check accumulating index op for RNIL before dereferencing it
ndb/src/kernel/error/ErrorReporter.cpp:
Include NDB version in ndb_X_error.log
ndb/src/ndbapi/NdbIndexOperation.cpp:
Let failed tcindxreq abort
ndb/src/ndbapi/ndberror.c:
New error code for out of index operation records
post review fix.
don't init hash value on restart operations
make sure that lock mode is correct in entire que.
ndb/test/ndbapi/testOperations.cpp:
Test case also for abort case.
(abort if not lock owner still fails!)
Redo abort handling according to descr. in Backup.txt
bug#9960 - ndb backup
increase wait completed timeout to 48 hours
ndb/include/kernel/signaldata/BackupImpl.hpp:
Add nodeid to reply to be able to fake reply during NF
ndb/include/kernel/signaldata/BackupSignalData.hpp:
new error codes
ndb/src/common/debugger/signaldata/BackupImpl.cpp:
fix printout
ndb/src/kernel/blocks/backup/Backup.cpp:
bug#9924 - ndb backup abort handling
Redo abort handling according to descr. in Backup.txt
ndb/src/kernel/blocks/backup/Backup.hpp:
bug#9924 - ndb backup abort handling
Redo abort handling according to descr. in Backup.txt
ndb/src/kernel/blocks/backup/Backup.txt:
bug#9924 - ndb backup abort handling
Redo abort handling according to descr. in Backup.txt
ndb/src/kernel/blocks/backup/BackupInit.cpp:
bug#9924 - ndb backup abort handling
Redo abort handling according to descr. in Backup.txt
ndb/src/kernel/blocks/cmvmi/Cmvmi.cpp:
Init own version
ndb/src/mgmapi/mgmapi.cpp:
bug#9960 - ndb backup
increase wait completed timeout to 48 hours
ndb/src/mgmsrv/MgmtSrvr.cpp:
Handle node failures activly
(mainly for backup...)
ndb/src/mgmsrv/MgmtSrvr.hpp:
Handle node failures activly
(mainly for backup...)
ndb/src/mgmsrv/MgmtSrvrGeneralSignalHandling.cpp:
Handle node failures activly
(mainly for backup...)
ndb/src/ndbapi/ndberror.c:
new error codes
ndb/test/ndbapi/testBackup.cpp:
fix return codes
ndb/test/run-test/daily-basic-tests.txt:
Add failure test cases to autotest
ndb/test/src/NdbBackup.cpp:
fix error codes
introduce checking of backup resources after each test
Make BUILDINDX RF_LOCAL aware
ndb/src/ndbapi/Ndbif.cpp:
allow TCKEY_FAILREF w/ invalid magic number also when NDB_NODROPPED_SIGNAL
ndb/test/ndbapi/testIndex.cpp:
Run only 1 update thread during NFNR2_O
Crash if ACC_CONTOPREQ was sent while ACC_LCPCONF was in job buffer
if ACC_LCPCONF would have arrived eariler (before TUP_LCPSTARTED)
operations could lockup.
But would be restarted on next LCP
-- LQH
1) Better check for LCP started that will also return true
if ACC or TUP already has completed
2) Remove incorrect if statement that prevented operations to
be started if ACC has completed
-- ACC
Make sure all ACC_CONTOPCONF are sent before releasing lcp record
i.e. use noOfLcpConf == 4 (2 ACC_LCPCONF + 2 ACC_CONTOPCONF)
Check for == 4 also when sending ACC_CONTOPCONF
ndb/src/kernel/blocks/dbacc/DbaccMain.cpp:
Make sure all ACC_CONTOPCONF are sent before releasing lcp record
i.e. use noOfLcpConf == 4 (2 ACC_LCPCONF + 2 ACC_CONTOPCONF)
Check for == 4 also when sending ACC_CONTOPCONF
ndb/src/kernel/blocks/dblqh/Dblqh.hpp:
Remove LCP_STARTED state
ndb/src/kernel/blocks/dblqh/DblqhMain.cpp:
1) Better check for LCP started that will also return true
if ACC or TUP already has completed
2) Remove incorrect if statement that prevented operations to
be started if ACC has completed
1) Make getNoParall into function instead of a procedure
2) Check for multiple transactions in "upgrade's" parallell queue
3) Set lock mode according to lock_owner's lockMode
NOTE: Does still not handle lock upgrade in case of aborts correctly
ndb/src/kernel/blocks/dbacc/Dbacc.hpp:
Make getNoParall into function instead of a procedure
ndb/src/kernel/blocks/dbacc/DbaccMain.cpp:
1) Make getNoParall into function instead of a procedure
2) Check for multiple transactions in "upgrade's" parallell queue
3) Set lock mode according to lock_owner's lockMode
Impl. stupid mgm lock against shutting down 2 nodes simultaniously
Add some more dump
ndb/src/kernel/blocks/dbdih/DbdihMain.cpp:
Impl. 2 new dump
1) to query nf handling state
2) that impl. sending of arbitrary signal...
to be used with extreme care
ndb/src/kernel/main.cpp:
enable critical and warnings in out file
ndb/src/mgmsrv/MgmtSrvr.cpp:
call handleStopReply on nfCompleteRep instead
ndb/src/mgmsrv/MgmtSrvr.hpp:
call handleStopReply on nfCompleteRep instead
ndb/include/ndbapi/NdbConnection.hpp:
release scan op of hupped trans at scan close to save memory
ndb/include/ndbapi/NdbScanOperation.hpp:
release scan op of hupped trans at scan close to save memory
ndb/src/ndbapi/NdbConnection.cpp:
release scan op of hupped trans at scan close to save memory
ndb/src/ndbapi/NdbResultSet.cpp:
release scan op of hupped trans at scan close to save memory
ndb/src/ndbapi/NdbScanOperation.cpp:
release scan op of hupped trans at scan close to save memory
ndb/tools/desc.cpp:
release scan op of hupped trans at scan close to save memory
ndb/src/kernel/blocks/dbacc/Dbacc.hpp:
set up long key page vars in ACC_LOCKREQ
ndb/src/kernel/blocks/dbacc/DbaccMain.cpp:
set up long key page vars in ACC_LOCKREQ
