Minutes of the ITC-Research Computing Standing Committee

Meeting on July 28, 2003

 

Members:  Alice, Bill, Brian, Dawn, Ed, Hamp, Kathy G., Katherine H., Jim, Mark S., Martha, Michael, Robin, Sue Ellen, Steve, Terry, Tim S., Tom S.

 

Attending: Alice, Bill, Ed, Hamp, Kathy G., Katherine H., Jim, Mark S., Martha, Robin, Terry, Tim S., Tom S.

 

Chair:  Tim T.  (Ed Hall chaired this meeting in Tim’s absence)

Recorder:  Alice

Click here to see the agenda from this meeting.

 

I.   Corrections to minutes from the last meeting on June 30, 2003

            None.

 

II.   Ongoing Discussion Topics:

1.  Aspen Cluster

·     Down time for upgrade and hard drive addition scheduled for August 5-6.  There was discussion about what exactly was to be included in the upgrade.  In the end, it was decided to keep it as close to Birch as possible.

·    How is the load?  Actually it is being used quite heavily – going to look at the PBS accounting logs to get a better idea of who is using the cluster.  We might consider implementing the recommendation for a “faculty guidance committee” (from the final report of the U.Va. Research Computing & Information Technology Task Force, March 2001) to set some guidelines/policies for allocating our high-performance computing resources – it has been suggested that the Academic Computing Advisory Council could perform this function.

2.  AIX SMP

·    Athena has hardware problems (will not boot) and is out of maintenance – so will need to find a substitute to install as the front end.

·    In order to upgrade the SMP box to AIX 5.1, we will first need to do some testing.  Likely that the upgrade can be scheduled by the end of September or sooner.

3.  E-School Linux Cluster

·    Cluster rollout has been hampered by a series of hardware failures.

·    They intend to make their setup/configuration as similar as they can to our clusters – but there will be differences.  These clusters will be used to explore grid computing.

4.  VPN and IP filtering for FlexLM license daemons

·    What we really need is authentication to FlexLM – but there really are no market factors that are pushing for this solution – actually vendors can potentially make more $$ if they do not solve this problem.

·    Could VPN work from home for Windows users?  Yes – but no guarantees.  So we can try using VPN while not promising that it will work well/easily.

·    Ed will talk to Matlab about a FlexLM solution.

5.  SP Migration

·    An e-mail notification about the migration has been sent to all SP users – stating that the system will be gone on 12/15/03.  Every contact with an SP user gets them a reminder to migrate.  A scan of the usage logs for the past 6 months yielded ~30 unique and active users -- of those, only 10 are heavy users.  We will contact them individually and get them to run on the SMP or on a linux cluster.

 

  

III.   New Topics:

1.  Incorporate unix lab machines as a cluster?  Ed and Katherine are looking at this.  Hamp reported that the Sun grid engine only runs on Solaris and linux – not on SGIs.  Katherine tried a tutorial – looks too difficult for users as they would need to write their own scripts to gather their own resources.  It might be possible to use PBSPro to manage resources including the lab machines – or to try Condor-G from the NSF Middleware Initiative.

 

2.  Tom asked when the firewall in Carruthers will be ready – as he has 2 CDR machines that he wants to put on the Carruthers network.  It will be the end of August at the earliest.  Also Tom wants to experiment with joining 2 lab machines with a cable from Crick –for the 6-8 people who run Gaussian in parallel.  Hamp will look at how difficult this might be to do.

 

Next meeting is September 29 – the August 25 meeting is cancelled due to Back-to-School.