WeeklyTelcon_20220705

Open MPI Weekly Telecon ---

• Dialup Info: (Do not post to public mailing list or public wiki)

Attendees (on Web-ex)

• Akshay Venkatesh (NVIDIA)
• Austen Lauria (IBM)
• Brendan Cunningham (Cornelis Networks)
• David Bernhold (ORNL)
• Geoffrey Paulsen (IBM)
• Hessam Mirsadeghi (UCX/nVidia)
• Howard Pritchard (LANL)
• Josh Fisher (Cornelis Networks)
• Josh Hursey (IBM)
• Todd Kordenbrock (Sandia)
• Tommy Janjusic (nVidia)

not there today (I keep this for easy cut-n-paste for future notes)

• Edgar Gabriel (UoH)
• Joseph Schuchart
• William Zhang (AWS)
• Artem Polyakov (nVidia)
• Aurelien Bouteiller (UTK)
• Brandon Yates (Intel)
• Brian Barrett (AWS)
• Charles Shereda (LLNL)
• Christoph Niethammer (HLRS)
• Erik Zeiske
• Geoffroy Vallee (ARM)
• George Bosilca (UTK)
• Harumi Kuno (HPE)
• Jeff Squyres (Cisco)
• Joshua Ladd (nVidia)
• Marisa Roman (Cornelius)
• Mark Allen (IBM)
• Matias Cabral (Intel)
• Matthew Dosanjh (Sandia)
• Michael Heinz (Cornelis Networks)
• Nathan Hjelm (Google)
• Noah Evans (Sandia)
• Raghu Raja (AWS)
• Ralph Castain (Intel)
• Sam Gutierrez (LLNL)10513
• Scott Breyer (Sandia?)
• Shintaro iwasaki
• Thomas Naughton (ORNL)
• Xin Zhao (nVidia)

---

v4.1.x

• v4.1.5
  • Schedule: targeting ~6 mon (Nov 1)
  • No driver on schedule yet.
• Looks like 5 PRs are reviewed and ready to be merged (https://github.com/open-mpi/ompi/pulls?q=is%3Apr+is%3Aopen+base%3Av4.1.x+)
  • William Would like merged PR#10513
• Brendan summarized 3 CVEs in libevent 2.0.x series, that are fixed in newer versions of libevent.
  • Discussed a bit on
  • Two of the CVEs are almost certainly NOT built when Open MPI builds libevent.
  • Third CVE is in some search function supporting search (haven't traced if we actually compile it in for Open MPI).
  • Fixed in libevent 2.1.8
  • Fixes DO cherry-pick to libevent embedded in v4.1.x
  • Perception problem, since CVEs are scary.
  • Found this with scans of Open MPI.
• Would not want to try upgrading libevent to v2.1.8 for v4.1.x
  • Fixes are contained to simple fixes (bounds/range checking fix)
  • Cherry-picks is much less risky
• Decided Brendan will PR the small cherry-picks needed to v4.1.x and try v4.0.x this week

v5.0.x

• Schedule:
  • Still Looking at end of summer
    • PRRTE (longest blocker task) is targeting late summer.
  • v5.0 RMs updated Project page of Critical Issues (https://github.com/open-mpi/ompi/projects/3) yesterday
• v5.0.x is on libevent 2.1.8 and isn't at risk for CVEs found against v4.1.x
• PR 10442 needs review for main for session fixes.
  • Biggest OMPI Issue 9869 - external deps on mpirun
    • No one actively working right now. Some iteration a while back.
  • Quite a few doc items
  • Newish issue regarding the partition communication features.
    • Since it's a new feature try to get these in as well.

PRRTE

• Ralph is deleting PRRTE v2.1 branch, and create a new release branch off of master
  • https://groups.google.com/g/pmix/c/6bT-mQq4Sqs/m/8Y0NsSbeBgAJ
  • Requires OMPIx v4.1.2
  • v5.0 RMs will discuss and see how this might affect schedule.

Main branch

• Continued bugfixing on UCX/OSC
  • Some should get backported to v4.1.x, but right now just v5.0.x
  • Passing MOST one-sided tests.
• Asked about atomics PRs (see below)
  • No discussion

Accelerator framework

• sm_cuda component was moved into framework.
  • nVidia is ready to do some testing.

Joseph posted two atomics PRs

• 10492 and link to 10487
  • Geoff STILL needs to test on ppc64le, will get to this TODAY!
  • No Discussion today.

MTT

Face-to-face

• Wiki for face to face: https://github.com/open-mpi/ompi/wiki/Meeting-2022
  • Might be better to do a half-day/day-long virtual working session.
    • Due to company's travel policies, and convenience.