v0.4.11

• Introduced a concept of global YARN resources which can be shared across applications. Kudos to @lceleban-dd for adding this.

v0.4.10

• Fixed release of YARN client resources when submitting the Akkeeper application to YARN cluster.

v0.4.9

• Fix usage of the default staging directory in case when a user that submitted application is different from a user that runs the application.
• Akka Cluster is no longer a mandatory part of Akkeeper and can be disabled by setting akkeeper.akka.use-akka-cluster to false.

v0.4.8

• Fix error handling when launching containers on YARN.
• Make sure that each instance of the Member Autodown Service has a unique name.

v0.4.7

• Fix split brain by terminating container instances if Akkeeper Master is unavailable during some configurable period of time.
• Fix more edge cases related to Akkeeper Master recovering from crash.

v0.4.6

• Refresh the list of instances in case when the local list doesn't match list in the storage.

v0.4.5

• Refresh the local list of instances after joining the cluster.

v0.4.4

• Allow Container Instance to proceed with its workflow when it's weakly joined the cluster.

v0.4.3

• Use Coordinated Shutdown when terminating a Container Instance.

v0.4.2

• Fix termination of the Akkeeper Master instance in case when Heartbeats are enabled.