unable to safely stop distributed data node

[Michael Dykman]

Hello all,

having got an apparently stable set-up of nbd-clusters  ( 2 machines 
each 2 datanodes, 1 api) I find that any attempt to stop any of the 4 
nodes is failng through hte  ndb_mgm console I am told

Executing STOP on node 4.
Shutdown failed.
*  2002: Stop failed
*        Node shutdown would cause system crash

I tried the hard way and $ kill'ed 1 data node which took out every 
componnt but the managerment server which suggests the message above is 
correct.

has anyone else come across anything like thing?  I'm including my 
config below in case there is something I've missed.

=====================================================================
[COMPUTER]
Id:1
ByteOrder: Little
HostName: abydos

[COMPUTER]
Id:2
ByteOrder: Little
HostName: giza

[MGM]
Id:1
ExecuteOnComputer: 1
                                                                              


[API]
Id: 2
ExecuteOnComputer: 1
                                                                              


[API]
Id: 3
ExecuteOnComputer: 2
                                                                              


[DB]
Id: 4
NoOfReplicas: 2
ExecuteOnComputer: 1
FileSystemPath: /var/mysql/ndb_data/node1
                                                                              


[DB]
Id: 5
NoOfReplicas: 2
ExecuteOnComputer: 1
FileSystemPath: /var/mysql/ndb_data/node2
                                                                              


[DB]
Id: 6
NoOfReplicas: 2
ExecuteOnComputer: 2
FileSystemPath: /var/mysql/ndb_data/node1
                                                                              


[DB]
Id: 7
NoOfReplicas: 2
ExecuteOnComputer: 2
FileSystemPath: /var/mysql/ndb_data/node2
                                                                              


[TCP DEFAULT]
PortNumber: 10002
                                                                              


#first data node to manager
[TCP]
NodeId1: 1
NodeId2: 4
                                                                              


#second data node to manager
[TCP]
NodeId1: 1
NodeId2: 5

#third data node to manager
[TCP]
NodeId1: 1
NodeId2: 6

#fourth data node to manager
[TCP]
NodeId1: 1
NodeId2: 7

# DN 1->2
[TCP]
NodeId1: 4
NodeId2: 5

# DN 1->3
[TCP]
NodeId1: 4
NodeId2: 6

# DN 1->4
[TCP]
NodeId1: 4
NodeId2: 7

# DN 2->3
[TCP]
NodeId1: 5
NodeId2: 6

# DN 2->4
[TCP]
NodeId1: 5
NodeId2: 7

# DN 3->4
[TCP]
NodeId1: 6
NodeId2: 7

# first API client to first data node
[TCP]
NodeId1: 2
NodeId2: 4

# first API client to second data node
[TCP]
NodeId1: 2
NodeId2: 5

# first API client to third data node
[TCP]
NodeId1: 2
NodeId2: 6

# first API client to fourth data node
[TCP]
NodeId1: 2
NodeId2: 7

# second API client to first data node
[TCP]
NodeId1: 3
NodeId2: 4

# second API client to second data node
[TCP]
NodeId1: 3
NodeId2: 5

# second API client to third data node
[TCP]
NodeId1: 3
NodeId2: 6

# second API client to fourth data node
[TCP]
NodeId1: 3
NodeId2: 7