While running a job an error message is returned with one of the following phrases:
- timeout while waiting for channel to be ready for read .org.apache.hadoop.net.SocketIOWithTimeout.doIO
- failed on socket timeout exception: java.net.SocketTimeoutException
This error can occur as of a communication timeout setting being exceeded during the execution of the job.
If the error is the result of a SocketIOWithTimeout exception then it will be necessary to increase the value of mapred.socket.timeout. If the error is the result of a SocketTimeoutException exception then it will be necessary to increase the values
- ipc.client.connect.timeout (symptom)
- mapred.job.tracker.handler.count (likely cause)