Losing connections to server nodes. (no replies)

Hello,

I have a cluster in Amazon with 2 data nodes and 2 server nodes.

There is a process which reads 10,000 records from one table, processes it, and then inserts the new data into another table.

This does it in python using 300 threads using the 2 server nodes with 300 max connections each.

The script iterates between the 2 sql nodes.

When I get to thread #500 aprox I start getting error messages from the nodes..

OperationalError: (2003, "Can't connect to MySQL server on '10.xx.xx.xx' (110)")

then

OperationalError: (1296, "Got error 4350 'Transaction already aborted' from NDBCLUSTER")

I checked the error logs in both server nodes and the NDB node but did not see any entry for that time frame.

Any idea what I should check?

Thanks