LWG Minutes 2016-10-19

From OpenSFS Wiki
Revision as of 07:12, 20 October 2016 by Jgmitter (talk | contribs) (Created page with "== Attendance == Cray: Cory Spitz <br /> ORNL: James Simmons, Dustin Leverman <br /> Indiana: Ken Rawlings <br /> Sandia: Ruth Klundt <br /> Intel: Joe Gmitter, Andreas Di...")
(diff) ← Older revision | Latest revision (diff) | Newer revision → (diff)
Jump to navigation Jump to search


Cray: Cory Spitz
ORNL: James Simmons, Dustin Leverman
Indiana: Ken Rawlings
Sandia: Ruth Klundt
Intel: Joe Gmitter, Andreas Dilger, Peter Jones


New Actions Captured:

  • None

Existing Open Actions:

  • None

Actions Recently Closed:

  • None


2.9 Release Status/Update

  • We currently have 4 blockers for the 2.9 release with 3 of the 4 having patches in flight.
  • The long pole remains to be closing out on is SSK development. We are waiting for the last patch to be submitted to gerrit that completes the changes suggested by Intel security experts. The final patch changes how key generation is done and improves the session key generation process significantly.
  • Is there any testing feedback from the community?
    • James: We have brought up a testbed with 2.9 servers and clients on RHEL 6.8. It is looking fine so far with no issues found.

2.10 Release Update

  • No significant news to report. Once we branch when reaching code freeze for 2.9, we can begin looking at 2.10 landings.
  • James: What are the big features planned for 2.10?
  • James: Will features such as Multirail LNet merge in patch by patch or will we be looking at merging the two branches together?
    • Andreas: The work is already split into patch series and we would be landing those on master as we have done in the past.

Upstream Lustre Client

  • The 4.9 merge window has closed and the 4.9 release will be up to 2.7.55, so a few of the 2.8 patches made it in before it closed.
  • We are still moving along and James has another 40+ patches needed that will get us up to ~2.7.60
  • Now that we are in 2.7+ area, we should really consider being more diligent in inspecting the upstream patches carefully. James suggests that we should have outside reviewers besides James and Oleg look at the patches.
    • Andreas: How often are issues found?
      • James: There has about 5 issues in ~1000 patches.
    • Andreas: Do you test before your send them?
      • James: Yes. The idea would be to have someone review again upstream to make sure it was ported correctly. This is a good way for people to contribute that are looking to help.
    • James: Andreas mentioned before Intel was setting up a test infrastructure to test patches for upstream compatibility.
      • Andreas: We have not made the staging area available to all yet as tests would fail due to other changes not being present on that branch. However, James has access to it and perhaps submitting a patch there to see what is still failing would be a good way to see what still needs to be done to get this complete.


  • The Mediawiki upgrade and migration work continues.

Other Business

  • Dustin: What technical content do we want to see at the openSFS BOF?
    • The thought is that the talk should focus on lustre installations for the next 5 years. Peter’s thoughts were to have high level overviews, spending a few minutes each, on ZFS, PFL, DNE, Multirail LNet
    • We want to have enough content to have a good discussion without the talk going dry. Are there any other suggestions or thoughts on the BOF discussion/content?
      • Andreas: It really depends on the goal of the BOF. If the goal is to solicit input as to what is a desirable feature, it would be worthwhile to extend the scope of the discussion to include potential features coming further down the road such as File Level Redundancy.

Next meeting will be on 2016-10-19