Skip to end of metadata
Go to start of metadata

Attendees:

Canonical Meeting Template (as a reminder):

1. Review of meeting agenda
2. Addressing issues

  • Description/Status of issue
  • Review of action items from previous meetings if any
  • Recommended actions and discussion of same
  • Close issue or identify next action items

3. Status updates from SMT members

  • Status update on significant activities
  • Review of prior open issues if any (and not addressed earlier in agenda)
  • Identification of new issues if any (and not addressed earlier in agenda)

Agenda/Notes

1. Review of Meeting Agenda: Any comments/additions? 


2. XSEDE2 Reporting (Leslie Froeschl

Previously:

  • Previous discussion archived: SMT Meeting - September 10, 2020
  • (9/24) No update
  • (10/8): IPR13 coming up starting first week of November. L3 text will be due 11/3 and L2 text 11/5. Please try to be on time!
  • (10/22): Reminder about IPR13 (see 10/8 reminder). Risk review went out late yesterday. Please review your risks by 10/31 at the latest. 
  • (11/5): L2 inputs for IPR13 due today!

    • Linda: difficult discussing some KPIs performance since we don't have in-person training. Do our best work in-person with students & faculty. Feel that we're losing touch with the community. 
      • John: Best we can do is just interpret what is going on as best we can. Will add note to summary about impacts of COVID19 on outreach efforts & other in-person efforts. Ok to qualify that we're speculating because we don't have experience in this space. Talk about strategies we're taking: how much we're doing virtual, assessment of virtual events/contacts. 
      • Trying think about strategies & mentioned that in report response. 
      • Kelly: some things have been surprisingly positive (esp student programs), but exhibiting is really difficult. Taking a toll on staff. 
      • Students in current programs were recruited last year. Now we should be recruiting for the rest of the plan year. Getting word of mouth from existing/past participants, but recruiting pipeline is 6-12 mos and we're not filling it. 
      • Some universities have stopped travel through May already.
  • (12/3) IPR13 was submitted on time. Process went very smooth. Thanks to all! 

Discussion:

Decisions:


3. Terminology Task Force (Linda Akli

Previously:

  • Previous discussion archived at SMT Meeting - September 24, 2020
  • (10/12) Re-added to agenda
  • (10/22) List is being sent out by email. Look at the list and provide any feedback (anything should be added/removed). Include source of where you identified that as a problematic term. Based on this list the team will start cleaning up website etc.
    • "White paper" is a term on the list. "Position paper" is a better option
    • Susan & team has done a great job!
  • (11/5) Thank you for input/comments. Task force is going back through spreadsheet to add column on context. Want to order terms to make sure those that would be most likely used by XSEDE appear at the top. Should bring back to SMT and then will lay out a plan for going through content to clean up terminology as needed. 

Discussion:

  • Open MP/MPI session at SC? John has not heard anything but will follow up with them. 
  • ACM has posted something on their site with about 5 terms. 
  • Will add link to Terminology Statement at bottom of workshop page. 
  • Boz involved with the group, so he will help disseminate info about this. 
  • Statement was sent to the group earlier this week. 
    • "While some third-party materials may be beyond our editorial control" sounds negative. Want to accept feedback whether in our control or not. Like to have it included so we acknowledge there are things beyond our control. Suggest replacing "While" with "We recognize..." 
    • Final suggestion: "We encourage you to reach out to... as we recognize that some third-party materials..."
  • Task force will present status/plans during quarterly meeting
  • "Words matter" is important. Might streamline: 

    Words matter. XSEDE's commitment to fostering and promoting an inclusive environment for all users, staff, and the wider community extends to all language and terminology all of our materials. We encourage you to reach out to the XSEDE Terminology Task Force with any feedback, recognizing that some third-party materials may be beyond our editorial control.

Decisions:


4. SPF L2 Application: University of Kentucky & Move from L2 to L1: Purdue Anvil (Ron Payne

Previously:

  • (11/25) Ron sent SMT the University of Kentucky & Purdue applications for review.

Discussion:

  • UK: Bob sent question (was sent to UK): The applicant writes: “All 5 nodes will be dedicated for XSEDE use/allocation with any unused resources being available for use by researchers at the University of Kentucky”.

    How exactly will this be done? In order for the nodes to be dedicated for XSEDE use, jobs launched by UK researchers will have to be preemptable. This could be a serious issue since some of the key large memory apps, such as de novo genome assembly and phylogenetic tree inference, can have long run times.

    • Only 5 nodes. Is it worth the effort? 
    • Large memory node is a specialized resource. 
    • 3TB nodes. Bridges2 covers large memory, but only have a handful. Had a lot of large mem nodes at PSC, but found many jobs could run on less. Had a bunch of big projects though that filled them up for a few months. Reduced number on Bridges2. Data driven modes of work are bursty. Need a lot for a short amount of time. 
    • Reluctant to add another resource, but would it be feasible to burst from Bridges2 to UK machine? Submit large mem requests, and they could run transparently on UK machine. Allocate through Bridges2. 
      • Doubt UK wants to be transparently behind Bridges. You enter XSEDE to be part of the national community. Guess everybody would like brand recognition. 
    • Award was made in 2016. Towards end of award. Understanding that this would be allocated by XSEDE but it hasn't so far. 
      • Can we even say no since NSF said it's to be allocated? 
    • They need to reply to questions.
    • At what level are we reviewing their operational policies/practices? If dedicating to XSEDE means everything needs to be preemptable... KY users at mercy of the few XSEDE users. There are some options. We should be careful about specifying how to operate the system.
    • Not a whole lot of info out there about their award. Appears to be 50 large memory nodes, so 5 is 10% of the nodes. Should clarify that. 2016 acquisition of MRI 1626364. Award states they will join XSEDE. 

      https://www.nsf.gov/awardsearch/showAward?AWD_ID=1626364&HistoricalAwards=false

      • What would be minimal amount of time we would consider for accepting? Minimum of 2 years. 

    • KY is EPSCOR state.
    • Don't want to begrudge anyone trying to get into the national landscape.
    • Need to make it work with them. 
    • Award ends last July 2020. Extension? How long will this be available?  
    • Fabric award: follow-on to Genie. Wanted to be a means of allowing researchers that are in XSEDE that have interest in access Fabric to have an easy gateway. 
    • Lifetime of machine at this point? 
    • How they support that machine since end date is past? 

  • Purdue Anvil (forwarded by Ron on 11/11)
    • Track 2 Cat 1 system 
    • Request new letter stating: Should be modified to reflect that Anvil ( not Purdue) becoming L1 while Purdue remains L2 provider. 
    • Should state how much of GPU & large memory resources will be made available through XSEDE

Decisions:


Around the call up

Follow this basic outline:

  • Status update on significant activities
  • Review of prior open issues if any (and not addressed earlier in agenda)
  • Identification of new issues if any (and not addressed earlier in agenda)
  • Identification of recent improvements implemented

Updates from SMT members:

  • PI (John/Ron) - 
    • Discussion about updated draft transition plan w/ request from NSF to have webpage with it linked. John pushed back on this. XAB was split on this. We will produce the doc and make available through IDOR, but concern about whether it might compromise competitiveness in solicitation. Less opposed if we: 1) state on webpage what we're providing are capabilities that are transferrable to awardee, but don't state priorities/appropriateness of capabilities. Will put things on the table but up to you to decide what to pick up. Like L2s to make pass through transition plan to look for anything that might compromise our competitiveness. Alleviates showing our cards too much, but satisfies NSF.  
      • Didn't think XAB was split. Felt they thought it an inappropriate ask from NSF
      • List of what could be transitioned doesn't seem to compromise us. Seems strange to put on a webpage. 
      • Nothing in initial draft plan that you couldn't glean from going through our wiki
      • Part of writing proposal is doing the legwork. We have advantage because we've been doing it for so long. We shouldn't have to do competitors legwork for them. 
      • Are there artifacts from project that NSF could say we funded so we must put on list. 
      • Provide links to wiki. Provide less rather than more. Make sure everything we list is already publicly available. 
      • PEP states that we will provide draft transition plan and then detailed plan to successor(s). 
      • Want to ensure user data is stewarded properly and transitioned carefully. 
      • Can tell you services we provide without saying what works well and what doesn't. Magic is in collection of people, how we focus, etc. Not telling anyone that. 
      • Could make available on the wiki, which is already public. 
      • Current draft plan includes most important services to be transitioned. This needs reviewed/updated. 
      • Big question is why we're being asked to make it so public. Doesn't feel that we're being treated like others. 
      • Can't imagine NSF would point to our doc and say this is what you need to do. What we say may not be what they'd want. 
      • More likely would say here is what is currently being done. 
      • In conversation with Bob, we were asked for webpage with schedule for production of updated plan and link to it when it's final. 
      • Could ask them to talk with us about how they're planning to use it. 
      • Point to list of all XSEDE resources & services. Doesn't provide explanation of XRAS
      • Put up org chart with links to all L2/3 pages on wiki. All details can be found there. Don't believe NSF would accept list of links as the draft transition plan 
      • Could put subsection on Project Documentation about draft transition plan. This page is already linked under About on xsede.org
  • Tim
    • Reminder to L2s that we have a session at the quarterly meeting to talk about the Interconnections Diagram and we're hoping to have input back from the L2 areas for that meeting. Email Tim with any questions or concerns!

  • CEE (Kelly) - 
    • No update
  • ECSS (Bob S/Phil/Sergiu) -
    • AMD training Dec. 8 at 12pm ET. Anyone is welcome to participate. Let Bob/Phil know. 
  • XCI (Dave L/Rich) -  No update

  • Ops (Greg/Victor) -   No update
  • RAS (Dave H/Ken/Nathan) -  No update

  • Program Office (Ron/Leslie) - 
    • Reminder: Enter Publications in XUP and documents in IDEALS
    • Reminder: Reference collaborations in tracking worksheet: https://docs.google.com/spreadsheets/d/19YRAPeAForh5I9Xvhj8nKjROe3-JN7a1caLYbNamPaw
    • Reminder: Address recommendations (Jira) 
    • L2s will be expected to respond to recommendations of climate study report during December quarterly meeting. Reports were sent by SP&E. 
    • Send Leslie nominations for 2 open XAB positions to increase diversity on the board.
    • Reminder: All L2s and L3s should be offering to their staff the option to request (or have their institutional supervisor request) feedback from their XSEDE manager on their XSEDE performance to be considered as part of their institutional evaluation. 
  • SP Forum (Ruth) -  

    •  No update
  • UAC (Emre) -   No update
  • NSF (Bob C) - 


Next Meeting:     SMT - December 17, 2020


  • No labels