Minutes of the ITC-Research Computing Standing Committee

Meeting on October 31, 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., Steve, Terry, Tim S., Tom S.

 

Chair:  Tim T.  Recorder:  Alice

Click here to see the agenda from this meeting.

 

I.    Corrections to minutes from the last meeting on September 29, 2003

            None.

 

II.   Ongoing Discussion Topics:

1.  Reminder that the SP is scheduled for retirement on 12/15/03

·     Will contact the active users (19 since June, and 22 users from 9/02 – 9/03).

2.  Faculty Advisory committee for Research Computing

·    Reminder:  ACAC is meeting conjointly with UCIT during this fall – our aim is to talk with them in the spring.

3.  Crick to be split in two with half behind the more secure network in Carruthers

·    Firewall is being tested next week

·    Crick to house patient data sets so has to be behind a firewall – but do not need the whole 4 processor machine for this – thus the split

·    Related item:  Tim’s group is looking for a policy statement for researchers about HIPAA, state regulations, etc.  Jim thinks we need to do the same as the hospital – Mark found a public HSC statement about HIPAA and privacy – also Shirley has a document about what HIPAA covers.

4.  E-school Linux cluster rollout progress report and setup/configuration

·    Need to get Jeff to come and give us an update.  Ed thinks they have some testers on it.  (Engineering has known all along that this would not be a quick installation/rollout.)

5.  VPN and IP filtering for FlexLM license daemons  Several vendors are now requiring it, not just ESRI.

·    Hamp can filter by license manager – PC and Mac users can use the VPN – and we could implement a dedicated license manager machine.  Should we explore other alternatives for linux users?

·    There was discussion about applying the filter for just those vendors that require it – or to do this for all vendors.

·    Also need to clarify what is the “UVA community”?  Would it include Mountain Lake? Blandy Farm, etc.?  What would happen to Continuing Education sites and students?  Only users in WHOIS can get a cert for the VPN.

6.  Clustering UnixLab machines

·    Katherine has been testing Globus on hera and nike and is having troubles with IP tables – Bill suggests logging.

·    On the horizon – what’s the future of the Unix lab?

        There was general discussion about why we might want to try to cluster the unixlab machines (to be “cutting edge”? to stay up with others who are experimenting with Globus, Condor-G?) – and to weigh our efforts against possible benefits.

        What else might we do with that space?  Will turn on console logging and see who’s using it now.


III.   New Topics:

7.  PBS Pro upgrade on Birch to get in version sync with Aspen

·    This will need a downtime – so wait until end of semester. In meantime, Bill can look at what’s really involved.

8.  Aspen Cluster:  When can 200GB disk be installed on the front-end? Replacement here? When to schedule downtime to install it?

·    Bill thinks that maybe it needs a better cable; downtime to add the disk is scheduled for 11/24.

9.  Ganglia/Gmond usage page switched over ResComp pages links.  Making “ganglia” graphics work for UnixLab computers?

·    Switched over and looks good so far.

10.  The “Digital Services ‘One-Stop’ Planning Committee” of Library and ITC folks is exploring how digital/technology support and services are offered in Library and ITC – report due before Christmas.

11.  Growing concern about how much downtime (planned and unplanned) Clusters have, how frequently there are glitches, problems, etc., that create problems or crashed jobs for users or extended downtimes.  How can we decrease these and manage expectations better in order to avoid the Clusters becoming afflicted with the “maggie syndrome”?

·    It’s a user credibility issue.  We knew that Aspen/AMD was not robust before we got it.

·    Bill said that most of these failures are traced to DIMMs failing – and Aspen replaces them with the same stuff.  So would Aspen give us better replacement DIMMS if we gave them more $$?  Would it be worth it to try to get all the memory upgraded?  Jim wants more details about the Intel memory.

12.  Tom reported that the ACHS and Research web pages are going to be merged.

 

 

Next meeting is Monday, November 24, at 10:30 AM.

<--- Go to: ITC Research Computing Standing Committee Home Page