Project lead: Oliver

Others: Alex, Sidhard (sp?), Boris, Liang (and in the future, anyone else in the group).

Goal

Get application on Skeeve (itself called Skeeve?) to run on contemporary, supported OS and hardware (ex. 64-bit).

Project status

4/5/13 (Fri): Skeeve has been removed from Joanne's office to a temporary storage spot in ChemIT's offices where it remains not plugged in.

To do's:

• Peter to confirm there are no salvageable parts in the computer so the computer can be disposed by ChemIT staff.
• Confirm that the research group currently has no plans to reinstall the source code on another computer.
• Jack approves that this project is to be considered completed.
• ChemIT staff properly dispose of this very old computer.. This includes ensuring the hard drive is destroyed so others can't get any data from it (ChemIT's standard operating procedure), to address Peter's concern.

Strategy

Short-term: Port to virtual machine (VM). Either with old OS or new OS, but either way, 32-bit.

Long-term: Convert via contemporary compiler to contemporary, supported OS and hardware (64-bit).

Option 1: CIT's VM service.

• Allow for "any old OS" (maintained by whom?), or only offering containers with OSes maintained by them?

Project notes and efforts

4/5/13 (Fri): Removed Skeeve from Joanne's office.

4/4/13 (Thur): Oliver spoke with Jack, and then with Liang, Alex, and Peter as a group. They were fine with the computer being removed and disposed of, especially considering its age. (Peter wanted to see if it had salvageable parts?)  Details:

• There would be no experimental data on that computer's hard disk since that data would instead be in each researcher's home directory (again, on CCMR's file share). This was confirmed by Barry, per Oliver's conversation with Barry earlier in the week.
• We confirmed that the source code was available in CCMR's file share. There are currently no plans to reinstall that code on another system.

4/3/13 (Wed): Removed Skeeve from power.

Late Feb or early Mar: Removed Skeeve from the network.

Oliver's meeting notes, 11/28/12's mtg

Barry:

• Depends on CCMR's AFS, which is also going away.
• Application was compiled to run on Skeeves server.
• Old 32-bit, on old hardware. Thought it might be RHEL~3 or so.

Ideas floated at meeting:

• On a new, fast Windows machine, create a Linux virtual machine (VM) with old OS version being used on Skeeve and port application over. Might port to new(er) OS.
• Key is to make this work with new file system so not depending on CCMR's AFS server/ services.
• Long-term: Must convert (re-write?) to make work with contemporary compiler, so can run on contemporary OSes and hardware. Until this happens, increasingly fragile.

Decision at meeting:

Instead of working on this now, focus on deciding on AFS provisioning, and on Eldore-related provisioning.

Past email threads, pre-meeting

Barry, 11/20/12, 2:34 PM

Skeeve is not only running 32 bit but also an old version of Linux, RedHat Enterprise 3. One idea to consider is installing a VM on Eldor with this version of Linux. Use NAT networking to secure the VM. There is also a question of the compiler. Is the skeeve code compiled with the old PGI compiler? If so, we need to figure out if this is a requirement and how to address it.

Peter, 11/20/12, 12:51 PM

Skeeve should simply be scrapped. We have to port 32-bit code to 64-bit platforms. The only issue is to find who will be in charge. May be Dave Schneider could give us advice. I am not sure whether the large v4 CAC cluster is 32-bit or 64 bit. If CAC has anything 32-bit, these can be used to run 32-bit code before it is recompiled for 64-bit platforms.

Oliver, 11/20/12, 10:23 AM

(2) Migrating function of Skeeve to Eldore (porting code) and decommissioning Skeeve.  (As wth Eldore, CAC's services may be a solution)