Versions Compared

Old Version 5

changes.mady.by.user Henry Fowler

Saved on

compared with

New Version 6

changes.mady.by.user Simon Horman

Saved on

Key

  • This line was added.
  • This line was removed.
  • Formatting was changed.

...

Simon Horman (Netronome)

Sukhdev Kapur (Juniper)

Andy Gospodarek (Broadcom)

Agenda

  • Release schedule for 5.1 - (Still looking for Nov 15 as dev complete?)
  • Patches from Netronome
    • Based on the governance call on Friday 10/26 it appears that there are no licensing concerns for TF project
    • Discussion on the impact of model of the proposal (dynamically loading a proprietary binary) as an open source community and voting. If voting yes review on gerrit will need to take place
  • TSC/TC Elections: voting now for two members -vs- voting now for all TSC
  • TF Security Team - Revisited
  • TF Developer Design Forums

...

  • Still looking for Nov 15 for dev complete.
  • Release 5.0.2 is delayed. Ed Ting requested the commit id's for 5.0.2
  • Discussion of Netronome's patch. 
    • Henry reviewed a diagram that tries to summarizes the proposal by Netronome.  See the google document at https://docs.google.com/presentation/d/1o4FC2uPphAgR2mu8YNWEGWe6Mz4yDYyjgvLqkb5c4eM/edit#slide=id.p8 , and the current snapshot of that document on the Wiki at https://wiki.tungsten.io/download/attachments/1409118/Offload%20Design%20Options%20-%20Oct%2030%202018%20ver.pptx?api=v2
    • There were still questions about the tradeoffs between supporting vendor-specific designs in the short term vs. retaining motivation for working vendor-independent solutions in sister industry groups (e.g., on the Linux TC Module and RTEflowRTE flow). 
    • Joseph raised questions about testing. 
    • Sukhdev asked for a list of the specific patches and the blueprint.  Can Netronome provide these please?  (see response below)
    • Sukhdev took the AI to have the key experts review the code before the next meeting and provide feedback. 
    • Paul raised the point that upstreaming the patches at this time (i.e., in time for Rel 5.1) raises some concerns, and asked whether there is a feasible option for this to initially be implemented by applying a patch to the upstreamed version.  And hence providing additional time for review.
    • Paul asked for vendors other than Netronome to review the code and identify areas where there might be alternate preferred solutions. 
    • Netronome code that needs review: https://review.opencontrail.org/#/c/47035/. Patch set 1 is the original work by Mellanox, patchsets 2 and 3 are the delta added by Netronome. We would like to get other NIC vendors to review this. This patch is also relevant: https://review.opencontrail.org/#/c/42850/.