Click here for the agenda from this meeting.
Tom introduced David Drake, 1/2 time in Physics Dept, 1/2 time with ACHS doing image processing.
TODO:> Tom, Ed, and David should get together to discuss image procesing/visualization. INOVS should be added to software list for consideration along with IDL & PDWave.

To Dos & Old Business: Numbers in parenthesis match the numbers on the To-Do Table of July 26, 1999 that we reviewed at this meeting.

Item 1: Research Computing platforms
Hamp: PBS is installed and working on Solaris 2.7, which it isn't suppose to be ported to yet. Seems to be fine, only had to make a very few changes. Thinks that installation on SGI is not far behind.
Ed still waiting on test account from Chip S.
Hamp: domain name for orange and teal cluster has been unilaterally and irrevocable set to .unix (not .unixlab). Needed to have separate domain name in order to use load balancing of interactive users. Accounts part is done. SP & SGI accounts are not a subset of unix. Will merge unixlab database with these mid-September.

Item 6: SP Upgrade
Hamp: Problems with SP turned out to have nothing to do with SSA array, it was a bug in the diagnostics, problem with new switch diagnostics that pointed to SSA as fault, but actually was something awry with the switch.
It's all been resolved. The SSA array is connected to the SP, half of them are on HSM.

Item 5: NPACI/Legion
Ed: No NPACI users. He has re-vamped NPACI webpages.
Started testing PBS on Grimshaw's Centurions in December. Recently heard that 64 Intel boxes running Linux of Grimshaw will be using PBS soon - will contact Grimshaw's group and see about test account.

Regarding Grimshaw's account offers of June 18; we agreed that we'd do testing on our setup and then contact him when we're ready. Not there yet.

Tom noted that folks are using Legion, both researchers here and elsewhere. Is a "legion-help" mail list that gets some traffic.

Item 2: HSM
Hamp is using and doing regular file dumps from NetApp to HSM, these are regularly larger that the 5 GB disk cache and the tapes keep up! This may also be a way to handle NetApp backups in future.

Tom reported that he's been transfering 40 GB to HSM recently, via FTP. HSM definitely kept up; did have some problems with FTP crashing/timing out.
Hamp: there's 3 tapes per backups, double tapes in robot and a backup (which may not be restoreable?) in user area.
Tim S.: will we have 2 file systems - one without a dual copy?
Discussion on this - Hamp, would only make sense with the "public" area, and it will be the smaller of the two areas (user and public). So may as well keep dual copy for both areas.

Space allocations: Hamp: currently no quotas or restricts on who may use.
Tim S. thinks it should be individuals and research groups.
Hamp would like to not allow generic account ids. Tie it to faculty members user-id and then have project directory names under that. Tim S: As for space allocations-he suggests 100 times blue.unix quota, with an automatic doubling of quota once upon request.

TODO> Tim T. draft space/quota policy and circulate, then Tim S. will take to directors for approval.

Tim S.: Need to contact David Seaman in Alderman Library about his HSM space needs, and need to contact Alderman in general about this (e.g., Special Collections).
Hamp: They truly need archival storage.

What about access methods? NFS only? FTP? SAMBA?
Tim S: thinks yes to all there as long as there's no interactive access allowed.
Tom: what about performance?
Hamp/Tim S: it's about like the Home Directory on a Macintosh.

Discussion of LongTmp in relation to HSM?
Hamp-move the LongTmp users to the HSM and delete from longtmp.
Dawn: Do we need LongTmp anymore?
Tim S: No, add LongTmp's space into HSM disk cache.
Migrate LongTmp users to the HSM as their two-week allocations run out.

Maggie to HSM migration
Hamp: Maggie to HSM migration proceeds. Have set Maggie to read-only, when user requests write, s/he gets moved to HSM and account setup there.
Have about half the users moved from Maggie. It's painfully slow to move them. The migration server is broken. Have stopped telling users to "let me know when I can delete Maggie files", am now saying "unless you let me know, I will be deleting Maggie files in a few days." Doing this because he needs to space to stage files in for migration.
Tim S./Hamp. will keep migrating as quickly as can- Hamp's not sure the migration server is Year2000 compliant; plan is to have it out of service before January 1.

Account creation/management
Hamp: Archive storage not in research password database. Research password database is subset of blue.unix, it's mirrored, and different physical database.
Tim S.: Will Marge/Norma be able to setup HSM directories in future?
Hamp: He's working on them being able to do so through their 'adduser' program. Account requests could be handled through same procedure as used for longtmp now. Should be e-mailed to group alias so not just one person getting account request.

Item 8: Research Computing Support Center
Dawn: Research Computign Support Center plans progressing well. It will be in 244 Wilson. Have cubical furniture configuration. Considering whether need to staff Small Hall; concentrate on Center in Wilson. Will need to discuss further.

Item 3: Research News
Tim T. did items for Office of VP for Research and Public Service. Will use this as basis for Fall Research Newsletter.

Item 9: Research Advisory Committees
Tim T: Should Research Computing Support Center have advisory board, something similar to ACHS's model or use ACAC as advisory board?
Tim S: UCIT is re-vamping committee structure, getting rid of little expert sub-committees. So should just use ACAC as advisory board.
ToDo:> Tim T. e-mail Catherine Kane and get on ACAC agenda.
What about UCIT recommendations report? Tom: UCIT hasn't met since April, as far as he knows report never released. He'll check with Melinda.

New Business

Side discussion about Steve Nock's request for 'www.marriagematters.edu' URL Robin, Dee, Hamp, Tim S.
Bottom Line: need to make Guidelines/Policy clear and readily available for both ITC Staff and users to consult.

Status of SUR Program
Ed asked about status of SUR program?
Tim S.: it's very restricted and targetted now mostly to business. He thinks it's like we won't get any funds this coming year. Engineering and Business school trying to partner to come up with an end-run around. Likely to be limited funds to upgrade the SP.
Tim S: better off buying a big Dell, 8 CPU servers and cluster them together on a Linux based OS.

Meeting adjourned arount 11:50, next meeting September 27, 1999, 10:30 AM

Go to: ITC Research Computing Committee Home Page

by Tim F.J. Tolson