Minutes for the ITC-Research Computing Standing Committee Meeting

August 29, 2005 at 10:30AM

 (in ITC-2015 Ivy Road, Room 102)

 

 

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

 

Attending:  Alice, Bill, Ed, Hamp, Katherine H., Jim, Martha, Nancy, Terry, Tim S., Joe S.

Chair: Tim T. Recorder: Alice Howard

I. Corrections to minutes from last meeting on June 27, 2005?  Under ‘Teak Cluster” should be G03 (not GL3; refers to Gaussian2003).

 

II. Ongoing Discussion Topics:

 

1.  Cluster projects:

·    “Condo” Cluster (aka cedar):

o        Have FIVE researchers participating: Leo Zhigilei (lz2n) – 8 nodes w/ ETF $; Jeff Shabanowitz/Dina Bai (Chemistry)- 8 nodes; Phil Parrish/Rich Gregory – 4 nodes; Matt Neurock – 4 nodes; new researcher, Avik Ghosh (av7rq), EE, 10 nodes (@Mitch’s request).

o        Outstanding issues:

§         Still having problems with insufficient power and A/C in the machine room – Facilities Management will be adding more but do not have a timeframe yet.  Meanwhile are able to keep up all but 4 nodes.

§         Ganglia reports from cedar – what Bill turned off made no difference – TimT will check with SteveL

§         AVAKI/Sybase Gateway node on Cedar?  Steve and Hamp are working on it, but it’s not yet set up.

§         PBS scheduler groups for “condo purchasers” – Katherine is working on them.

§         PBSPro priority queuing for paying participants and usage reporting are next on Katherine’s list.

o        New (fourth) cluster (2nd Condo Cluster) will go in manned machine room, per Jim.

·    Aspen & Birch cluster:  Upgrade done.

o        All but 2 of Aspen’s nodes are still functioning.

o        Maintenance is gone on Aspen and will be gone on Birch in December, 2005.

·    All the Linux clusters:  Users must change password on blue.unix not on cluster front-end; how do we notify users of this?  Could we change most frequently used password commands (e.g. passwd) to invoke message that users need to change password on blue.unix, not on front-end.

o        Decided to put an alias in the shells.

·    Oak Cluster:  timeline for adding the modules software that would make it seamless for IMSL and Java users to move from aspen to oak and make it easier for users to access different versions of the Solaris compilers.

o        Chip is going to look at it.

·    Teak cluster:  OK.

 

2.  /longtmp

·    Timeline for script to automate checking usage and reminding users – Hamp and TimT have not yet met

3.  Linux license server (cluster of three lm1., lm2., lm3.license).

·    Have moved as many research software licenses off of aix.license as can move off of aix.

 

4.  blue.unix upgrade of nodes to new 64-bit nodes and new version of OS. ITC-Transitions CDP is tracking as well.  Timeline?  We need to test software carefully.

·    Bill needs an HMC console in order to manage the machines – are there any other software packages on blue that were tied to a particular node (Lisrel was)?  Maybe Economic’s Gauss.

·    Katherine has found some issues with ANSYS and IDL.

·    Need to communicate to users going from 16 nodes to 3 nodes and 32bit to 64bit and ways this impacts them.

 

5.  Ganglia web pages:  Need to add cedar.itc.

·    Can we get  teak and oak clusters separated out from the general research computing cluster so we can see these nodes easily.  Is it possible to add athena.itc (SMP)?  TimT will check with SteveL.

 

III. New Business:

 

6.   Home Directory upgrade on Sept. 3 impacts on all Unix systems.  Will user quotas increase?

·    Quotas will increase – could go to .5 Gig with 1 double – or just go to 1 Gig and no double.

 

7.  Want to set up the RCSC Unix boxes as a resource group, rather like the “research” computing resources.  We don’t need them to be NIS clients, but we would like some way to grant a user access as a block to our machines and later remove that access, something like is done using www.people.virginia.edu, and some method of getting the password information that is a bit more efficient than manual cut and paste.

·    Can make them NIS clients – OR

·    Use vipw (but have to maintain a separate list to allow for differences between systems and a “users allowed” list).

 

8.  Ed working with Anita Jones and Andrew Grimshaw on possible one-day seminar on scientific computing over fall break (10/3) and longer course for J-term.

·    Possibly a one-day workshop on “high throughput computing” and a 2 week course on high performance computing during the J-term.

 

9.  Andrew Grimshaw and John Karpovich are looking for a linux front-end for their campus grid.

·    We can find something but need to understand their needs and process better – Hamp will follow‑up.

 

10.    Bill P. mentioned still having problem with users running intensive job on front-ends of clusters. Agreed that he’d send user-ids of users who sees who are doing this, and Ed or Katherine will contact to and find out why – find more appropriate resource, such as workstations at RCSC or how to correctly submit to batch mode.

 

Next Scheduled Meeting of this group: Monday, October 31, 10:30 AM in ITC-2015 Ivy Road, Room #102.

 

Next Scheduled Meeting of ITC ResComp Management team: Monday, September 26, 10:30 AM in ITC-2015 Ivy Road, Room #220.