This is an umbrella jira, like HDFS-1064, for discussing and providing references to jobtracker availability jiras (eg from JT restart on a host or to cross host fail-over).
NN Availability - umbrella Jira
TaskTrackers never (re)connect back to the JobTracker if the JobTracker node/machine is changed
Fault tolerant Hadoop Job Tracker
High Availability support for Hadoop
High Availability for JobTracker
[Umbrella] ResourceManager (RM) Fail-over