While running a job an error message is returned with the following phrase:
java.io.IOException: Bad connect ack with firstBadLink
This error occurs as the result of the loss of AWS EC2 Spot Node Instance Loss at the same time the job is sumitted and started.
The correct action here is to retry the query - this can be configured to occur automatically and increasing the value of dfs.client.block.write.retries will trigger more retry attempts. The Qubole Engineering Team is aware of the issue and investigating the potential to implement a fix for this corner case.