(MySQL Cluster) Error 2327: Memory allocation failure

[Consul]

Hi everyone. I hope this is the right place to ask this question.

We're trying to bring up a new node in our cluster to replace one on another computer which has become unstable. The new node is a Gentoo machine, and all of the others are Slackware 8 (we're trying to migrate over to all Gentoo - cool, no?). After bringing the other computer down and swapping out IP addresses, we try to bring up the new node, only to constantly run into this error:

-----
Node 3: Forced node shutdown completed. Occured during startphase 0. Initiated by signal 0. Caused by error 2327: 'Memory allocation failure, please decrease some configuration parameters(Configuration error). Permanent error, external action needed'.
-----

We're running MySQL 5.0.18 across four machines. This is our basic configuration (nodes 3 and 9 are the machine we are trying to replace):

-----
ndb_mgm> show
Cluster Configuration
---------------------
[ndbd(NDB)] 4 node(s)
id=3 (not connected, accepting connect from x.x.x.58)
id=4 @x.x.x.59 (Version: 5.0.18, Nodegroup: 0, Master)
id=5 @x.x.x.60 (Version: 5.0.18, Nodegroup: 1)
id=6 @x.x.x.61 (Version: 5.0.18, Nodegroup: 1)

[ndb_mgmd(MGM)] 1 node(s)
id=1 @x.x.x.57 (Version: 5.0.18)

[mysqld(API)] 4 node(s)
id=9 (not connected, accepting connect from x.x.x.58)
id=10 @x.x.x.59 (Version: 5.0.18)
id=11 @x.x.x.11 (Version: 5.0.18)
id=12 @x.x.x.2 (Version: 5.0.18)

ndb_mgm> all status
Node 3: not connected
Node 4: started (Version 5.0.18)
Node 5: started (Version 5.0.18)
Node 6: started (Version 5.0.18)
-----

Now, with MySQL cluster, each node has settings on how much memory can be taken up for tables and for indexes (two separate settings). Right now, on the nodes that work (the Slackware ones), these settings are 768MB and 512MB, respectively. If we decrease these values on the new node (the Gentoo one) to 512MB and 384MB, we can get it to come up, but it won't have enough memory allocated to hold our databases. (Each machine has 4GB of memory.)

I remember once running into some docs about a Linux kernel setting which determined the maximum size of a single file that can be opened in memory, and that this needs to be increased in order for the database to grow larger. This is so hazy in my mind I'm beginning to wonder if I dreamed it.

I have so far tried to ask this in several MySQL-related forums, to no avail. It's almost as if we're the first to ever see this happen.

Thank you all very much for any assistance you can provide.
_________________
Mother, please, is it just a disease that has them breaking all of my laws?
Check if you can disconnect the effect, and I'll go after the cause.
-- Moribund the Burgermeister

[Consul]

Sorry about the bump. Here is some more information, including part of the log dump.

Here's the output from the management node log when I try to start the NDB process on the new Gentoo server:

-----
2006-02-20 11:20:50 [MgmSrvr] INFO -- Mgmt server state: nodeid 3 reserved for ip x.x.x.58, m_reserved_nodes 0000000000001e0a.
2006-02-20 11:20:51 [MgmSrvr] INFO -- Node 1: Node 3 Connected
2006-02-20 11:20:51 [MgmSrvr] INFO -- Mgmt server state: nodeid 3 freed, m_reserved_nodes 0000000000001e02.
2006-02-20 11:20:52 [MgmSrvr] ALERT -- Node 3: Forced node shutdown completed. Occured during startphase 0. Initiated by signal 0. Caused by error 2327: 'Memory allocation failure, please decrease some configuration parameters(Configuration error). Permanent error, external action needed'.
-----

And here's the output from the startup command on the Gentoo machine itself:

-----
# ndbd-500 --initial --nodaemon
2006-02-20 11:20:51 [ndbd] INFO -- Angel pid: 24360 ndb pid: 14013
2006-02-20 11:20:51 [ndbd] INFO -- NDB Cluster -- DB node 3
2006-02-20 11:20:51 [ndbd] INFO -- Version 5.0.18 --
2006-02-20 11:20:51 [ndbd] INFO -- Configuration fetched at x.x.x.57 port 1186
2006-02-20 11:20:51 [ndbd] INFO -- Start initiated (version 5.0.18)
2006-02-20 11:20:51 [ndbd] INFO -- Error handler startup shutting down system
2006-02-20 11:20:52 [ndbd] INFO -- Error handler shutdown completed - exiting
2006-02-20 11:20:52 [ndbd] INFO -- Angel received ndbd startup failure count 1.
2006-02-20 11:20:52 [ndbd] ALERT -- Node 3: Forced node shutdown completed. Occured during startphase 0. Initiated by signal 0. Caused by error 2327: 'Memory allocation failure, please decrease some configuration parameters(Configuration error). Permanent error, external action needed'.
-----

Again, thank you all for your help.
_________________
Mother, please, is it just a disease that has them breaking all of my laws?
Check if you can disconnect the effect, and I'll go after the cause.
-- Moribund the Burgermeister