LWG Minutes 2019-05-30

From OpenSFS Wiki
Jump to navigation Jump to search


Cray: Cory Spitz
ORNL: James Simmons, Dustin Leverman
HPE: Olaf Weber, Christopher Voltz, Jeff Garlough
SuperMicro: Abe Asraoui
Sandia: Ruth Klundt
TACC: Ari Martinez
Whamcloud: Joe Gmitter, Peter Jones, James Nunez, Andreas Dilger, Patrick Farrell


New Actions Captured:

  • None

Existing Open Actions:

  • None

Actions Recently Closed:

  • None


LUG Developer Day

  • Is there any feedback on the LUG developer day?
    • Several comments that it was good and ran smoothly.
    • Dustin: It would be nice to add a more virtual presence for those not able to attend in person.
      • Peter: I have been in situations like this and being the remote person is very hard to follow along due to microphone pick ups being far away from commentary. It makes it very hard to follow along in discussions. Also, there are timezones to consider, etc…. We could try that in the future, however, we it is very much dependent upon a room configuration and the end value of it.

2.13.0 Release Update

  • We are gearing up for feature landing for MR routing and overstriping. PCC is in a topic branch and pretty close to having the merge commit. Also, the DNE work under LU-11213 looks just about ready to land.
  • UDSP hasn't had much review, is there anyone that has been testing UDSP that we are not aware of?
    • Cory: It wasn't in our plan to test UDSP before code freeze for 2.13.
    • It seems likely that UDSP will push to 2.14.
  • LNet sysfs is also having a number of issues that are being worked, however, it seems a bit risky with stability to include at this stage.
  • Self-Extending Layouts (Cory) - I don’t have details about test failures, but I know that Patrick and Vitaly have talked on a few occasions.
    • Patrick: The test failure situation is a little fussy and some are failing on zfs. Is anyone else reviewing the failures?
      • Cory: I have asked shadow to take a look at them but he hasn't finished going through all of them yet.
      • Peter: Any idea when we might see him get through all them?
      • Cory: We are not sure, will have to look into it.
    • Peter: Does the feature have a test plan and documentation ready?
      • Cory: Docs are a sore spot right now. There is a bit of a test plan and I have asked Vitaly to put it into the ticket. We could be a few weeks out yet have all the docs ready. The test plan is basically the PFL test plan, so not much change there. We will get something posted soon for discussion. User documentation is going to have to be after feature freeze in order to get it done.
      • Peter: Suggest that Cory comes up with a likely timeline for everything and then come back to see if it is worth sticking to 2.13 or deferring to 2.14.
  • Is there any thoughts on zfs 0.8 now that it is GA?
  • Cory: It would be worth to try to check it out since it is GA and if things go south we can always fall back. On the RedHat side, we should look at RHEL 8 more aggressively.
  • James: Has anyone tried zfs 0.8 to see how stable it is?
    • Peter: We have been running regression tests, but do we make it the default would be the question. Let’s confirm we can get equivalent test results to 0.7.13 and then go from there.
    • James: We have run into a bunch of 0.7.13 issues that have been resolved in 0.8.0.

Upstream Lustre Client

  • I have updated Neil’s patch set.
  • I have also tried the Whamcloud Lustre testing branch for upstream and it builds in 6-8 hours. We should look at how to make it cleaner. No tests were run though, so we also need to investigate there.
  • Neil has pushed the jobid suggestion as a new Jira ticket.


  • They have been doing updates for the maintenance releases.
  • Will put links to LUG presentations up and continue working on support matrix link.

Other Business

  • None

Next meeting will be on 2019-06-13 at 11:00am Pacific