Need some help to know root cause of this issue.
In this case the source host (Appmaster) was unable to received any response from destination as this is a faulty routing issue and packet was just lost. I checked the source code for the package "org.apache.hadoop.ipc.Client" and found that it uses PING utility to check response from destination host and keep trying until it received any responce. This is clearly mentioned in the this link http://grepcode.com/file/repo1.maven.org/maven2/com.ning/metrics.action/0.2.0/org/apache/hadoop/ipc/... "This class sends a ping to the remote side when timeout on reading. If no failure is detected, it retries until at least a byte is read." So, due to routing issue and it keep trying until the job was killed.
thanks to grepcode.com for facilitating easy way to read source code.