MySQL Bugs: #40682: All tables in used database deleted

Bug #40682	All tables in used database deleted - ndbcluster
Submitted:	12 Nov 2008 22:19	Modified:	19 Mar 2009 10:55
Reporter:	Michal Vrabel	Email Updates:
Status:	No Feedback	Impact on me:	None
Category:	MySQL Cluster: Disk Data	Severity:	S1 (Critical)
Version:	5.1.27-ndb-6.3.17	OS:	Linux (CentOS, Debian Lenny)
Assigned to:	Assigned Account	CPU Architecture:	Any

Description:
I am importing data by php script from myisam source on server A to Cluster solution (4x data node, 1x sql node) by inserts looks like:
insert into table (x,y,z) values (a,b,c),(d,e,f)......1000 values.
Tables has 10 - 300MB in myisam source (on different server).

When i am looking by phpmyadmin or by "show tables" in mysql linux console, I get almost list of tables, but somewhere this query is delayed and all tables with engine ndbcluster are lost.

This problem caused everytime, when I'll try import data once more. If I do not preview importing tables, import will be successfully done.

I have tried it with tables in tablespaces and tables not in tablespace.

Table definition with tablespace has STORAGE DISK definition.

How to repeat:
Import a lot of data and try to preview tables by "show tables" during import.

[NDBD DEFAULT]
NoOfReplicas: 2
DataDir: /SQLDATA

DataMemory: 3000M
IndexMemory: 1024M
BackupMemory: 64M

MaxNoOfConcurrentOperations: 100000
MaxNoOfLocalOperations: 110000
RedoBuffer: 16M
MaxNoOfOrderedIndexes=1024
MaxNoOfConcurrentIndexOperations: 16384
MaxNoOfConcurrentScans: 300
ODirect: 1
LockPagesInMainMemory: 0
StopOnError: 1
Diskless: 0

[MGM DEFAULT]
PortNumber: 1186
DataDir: /var/lib/mysql.mgmd

[NDB_MGMD]
Id: 1
HostName: cmdmgm1.cluster
ArbitrationRank: 1

[NDB_MGMD]
Id: 7
HostName: cmdmgm2.cluster
ArbitrationRank: 2

[NDBD]
Id: 3
HostName: cmddn1.cluster

[NDBD]
Id: 4
HostName: cmddn2.cluster

[NDBD]
Id: 5
HostName: cmddn3.cluster

[NDBD]
Id: 6
HostName: cmddn4.cluster

[MYSQLD]
Id: 2
BatchByteSize: 1M
BatchSize:800
MaxScanBatchSize:8M

[API]
Id: 8

[tcp default]
SendBufferMemory: 2M
PortNumber= 63132

Suggested fix:
Anywhere to save data in the cluster.

Severity change to Critical, because of lost data by system.

With version 5.0.61-community the same import caused:
Lost connection during query.
SQL node is not available for cca 3seconds and then is automaticaly up. 
No data has been deleted. This error is caused in the same "time" as ndb delete tables.

version of 5.x is of course 5.0.67, .61 is mistake.

this looks very much like bug#40854
how ever, there is not enough information in this bug report
to make any clear call, neither to make an attempt to reproduce.

No feedback was provided for this bug for over a month, so it is
being suspended automatically. If you are able to provide the
information that was originally requested, please do so and change
the status of the bug back to "Open".