Status update for task 6: Emulator user interface integration

[dga@cs.cmu.edu]

This milestone is three weeks late. Todo:

• Update with current completion status of this milestone
• Update with revised estimate for completion;

[dga@cs.cmu.edu]

Estimated completion date: May 14th. Putting these changes online will be too disruptive to active users; we need to wait until the end of the semester before we can make them live.

Major todo items:

• Update on consistency checking functionality: Coming soon.
• Moving Emulator control node functionality to a CMULab boss-controlled node to integrate experiment swap-in

Current completion status: 10%. Identified and separated functionality that needs to be moved to the ProtoGENI code from the Emulator specific code.

[pgunn@cs.cmu.edu]

We are currently in active development of this, having modified the CMULab client code to appropriately handle a static server (we wrote some nice documentation on how to do this), brought up said node, and modified our FPGA control software for running in that environment. The consistency checking functionality is, as far as I know, complete.

The remaining work to do is: Finish debugging problems with the FPGA software changes (1 week or less), integrate swap-in (1 week).

[pgunn@cs.cmu.edu]

Update: We had to do reasonable refactoring and adjustment to the FPGA-management environment on our emucontrol-1 node in order to migrate it to emucontrol-new, our emulab-managed "infrastructure-but-group-allocated" node. We also had to make modifications to the emulab client images to create this new type of node - these changes will be the foundation for further discussion with Utah regarding the correct way to have this style of node in the general case, as while the present solution will work with local emulab users, we are unsure how it will react with federated users and there are some security and data persistance issues that need to be addressed (I am considering making a new milestone for this work, although I am uncertain if it will be us or Utah to do that).

The simple solution is largely done, and we are migrating users to the new environment now. Consistency checking is already done, using XMLRPC to retrieve experiment data from boss and use as a template for FPGA programming. When the move is complete, we will be ready to have that software automatically invoke the FPGA-control software on the user's behalf.

[pgunn@cs.cmu.edu]

Update: New environment migration was delayed by some unrelated problems with the emulator, but is done now. The path is now clear to automation and it's under active development.

[Aaron Falk]

Dave will develop a more detailed plan on how to resolve this topic by Aug 21.

[pgunn@cs.cmu.edu]

The final step of integration is delayed by a lack of hooks in the Emulab event system for swap-in and swap-out events. We are presently looking into adding hooks for this; the event system is one of the more complex and least-understood part of emulab. We will also probably need to extend the ns parsing component for this (those changes should be small). Once we have hooks implemented and functional, this will be done.

[Aaron Falk]

Closing this ticket as 'wontfix' in Spiral 1. Justification can be found in ​http://groups.geni.net/geni/wiki/CmuLab-2Q09-status#A.Activitiesandfindings. Work to complete this integration appears in this projects Spiral 2 SoW.