Out of memory in NDB kernel and Table is full.

[Toan. Dang Anh]

Our cluster have 2 managements, 2 APIs and four nodes.
Each node have 10GB RAM,
We have two databases with total more than 8GB size.
Our configure file:
[NDBD DEFAULT]
NoOfReplicas=2
HeartbeatIntervalDbDb=600
HeartbeatIntervalDbApi=600
HostName=10.30.13.46
DataDir=/var/lib/mysql/mysql-cluster
BackupDataDir=/var/lib/backup
DataMemory=9G
IndexMemory=1G
TransactionDeadlockDetectionTimeout=12000
MaxNoOfConcurrentTransactions = 10000
MaxNoOfConcurrentOperations = 250000
TimeBetweenWatchDogCheck = 20000
MaxNoOfOrderedIndexes = 17000
MaxNoOfTables = 3500
MaxNoOfAttributes = 15000
NoOfFragmentLogFiles = 300
RedoBuffer = 1024M
TimeBetweenLocalCheckpoints = 10
TimeBetweenGlobalCheckpoints = 500
NoOfDiskPagesToDiskAfterRestartTUP = 80
NoOfDiskPagesToDiskDuringRestartTUP = 80
NoOfDiskPagesToDiskAfterRestartACC = 40
NoOfDiskPagesToDiskDuringRestartACC = 40
ArbitrationTimeout = 2000
[MYSQLD DEFAULT]
[NDB_MGMD DEFAULT]
[TCP DEFAULT]
# Section for the cluster management node
SendBufferMemory=1024K
ReceiveBufferMemory=1024K
[NDB_MGMD]
# IP address of the management node (this system)
Id=1
HostName=10.30.13.45
[NDB_MGMD]
Id=2
HostName=10.30.13.55

I always get the error out of memory and table is full.
Please help me solve this problem. I do not know is there any problem in
my config file?

Sometime, I can not restart our node on cluster and lose data. When this
happen, I should restart by ndbd --initial and restore from backup file.

Mysql cluster is very terrible with me. I always got problems.
Please give me your advice.
Thanks so musch.
Toan Dang Anh
VINAGAME JSC 
459B Nguyen Dinh Chieu, district 3, HCMC, VietNam
------------------------------------------------------------------------
-- 
Email: toanda@stripped 
http://www.vinagame.com.vn 
------------------------------------------------------------------------
--