Description:
Some changes, like changes to the TCP transporter, need a secondary restart of the management node. This is mentioned in the cluster logs.
The configuration is however loaded, but to avoid lots of disconnections, it was not activated (or so..).
..
Failed to configure TCP Transporter to node 12
Failed to configure TCP Transporter to node 1
2009-12-15 10:59:20 [MgmtSrvr] WARNING -- Could not reconfigure everything online, this node need a restart
2009-12-15 10:59:20 [MgmtSrvr] INFO -- Configuration 2 commited
2009-12-15 10:59:20 [MgmtSrvr] INFO -- Config change completed! New generation: 2
2009-12-15 10:59:30 [MgmtSrvr] INFO -- Node 2: Node 1 Connected
2009-12-15 10:59:30 [MgmtSrvr] INFO -- Node 1 connected
How to repeat:
Change the [TCP] sections, like the PortNumber, and reload the configuration of the management node, i.e. ndb_mgmd --reload.
Suggested fix:
Prevent need of restart, somehow?