I am having the same issue with slow HDFS.
I managed to resolve the second error:
java.net.SocketTimeoutException: 60000 millis timeout while waiting for channel to be ready for read (...)
by increasing DataNode Java xmx.
Hope this helps.
Well, actually we've fixed that issue. This was due to our network MTU, nothing to do with the FS. Our experimental MTU was far too-large, and was causing such behaviors. Getting back to the classical values bring things back to normal.