Roadmap
-
Basic operation of the cluster means that the cluster can be brought up and tasks can be run on it. This includes proper queuing and control of the tasks running on the cluster.
Note: Much of this work was completed prior to the trac instance being created. The completed tasks were never entered as tickets. The tracker only reflects work that was remaining after a large portion of the basic funcionality was already implemented.
-
Pydra is by definition, a remote code execution platform. It requires very strong security to prevent malicious users from gaining control of the cluster and running arbitrary code on it.
-
The cluster must be able to recover from failures in the cluster as best as possible. Fault tolerance ensures that a hardware or network failure will not cause a catastrophic failure
-
One of the core features of pydra is a management interface for both managing the cluster and the tasks on the cluster. This milestone is for producing the initial set of functionality in that user interface
-
A lot of code was written without unittesting. This is a complex piece of software so it is hard to verify it is working correctly. Unit tests for the initial code are needed to allow further development to proceed with less risk of breaking existing features
