Hi David,

There was a runner with a different versioned image, so it did not test correctly against the reference ABI.
We reset the image on that runner to the correct version, and now the tarball passes all ABI tests as seen here https://lab.dpdk.org/results/dashboard/tarballs/16879/. There should also be a new report sent out.

Thanks,
Brandon



On Thu, Sep 23, 2021 at 12:04 PM Lincoln Lavoie <lylavoie@iol.unh.edu> wrote:
DPDK Community CI Meeting Minutes
September 23, 2021

##########################################################################
Attendees

1. Lincoln Lavoie
2. Ashley Weltz
3. David Marchand
4. Ali Alnubani
5. Michael Santana

##########################################################################
Agenda

1. CI Status
2. Test Development
3. Any other business

##########################################################################
Minutes

==========================================================================
General

* Setting up DPDK Slack Workspace, use the following link to join: https://join.slack.com/t/dpdkproject/shared_invite/zt-v6c9ef5z-FqgOAS7BDAYqev_a~pbXdw
CI Status

--------------------------------------------------------------------------
UNH-IOL Community Lab

* Ready to deploy an internal system to periodically refresh Docker images
   * This will also update the ABI references to prevent false failures
* Refactoring Jenkins pipelines to increase efficiency of testing
   * We are actively working on reducing the backlog by optimizing allocation of resources within Jenkins
* Bare-metal downtime last week due to:
   * TRex ignoring return values indicating an error
      * This causes segfaults on any card with an xstats count > 0
      * It took a substantial amount of time to track this down
      * Upstream patch generated to TRex: https://github.com/cisco-system-traffic-generator/trex-core/pull/761
   * Changes to DTS since the last update requiring re-testing
* Bare-metal status
   * Mellanox and Broadcom on ARM are non-functional. All other NICs should be reporting up.
   * Mellanox systems updated to the latest versions of TRex (bug above didn’t impact that system), other systems were rolled back to known working versions.
   * Mellanox pass/fail criteria updated to 1.5%, per their request.
   * TRex and a couple of DTS bugs (divide by zero was reintroduced) will need to be patched, before upgrades can proceed.
* FreeBSD 13.0
   * Fix for 1 of the two unit test issues has been submitted: http://patches.dpdk.org/project/dpdk/patch/20210917150917.90892-1-bruce.richardson@intel.com/
* ABI failure 20.11
   * Need to investigate / confirm ABI failure for https://lab.dpdk.org/results/dashboard/tarballs/16879/

--------------------------------------------------------------------------
Intel Lab

* pf_smoke / test_pf_rss test case has been disabled, while waiting for a fix to be merged into the mainline.  Fixed has been merged into next-net, and should be in mainline by the end of the week.  Need to make sure we notify the Intel lab team to re-enable the test.

--------------------------------------------------------------------------
Github Actions & OBS

* OBS working on the final revisions for the patch.
* Github Actions and zero-day bot is currently down.  Server hosting the service had a hardware failure (occurred Monday).  Should be restored by next week.  The team has backups of the VM running the jobs.

==========================================================================
Test Development

* Finished categorizing the feedback DTS.  Creating an Excel (Google Sheet) issue tracker, so developers can be assigned, this may then be ported into Bugzilla
* Next DTS Meeting: TBD

==========================================================================
Any other business

* Next Meeting: October 7, 2021

--
Lincoln Lavoie
Principal Engineer, Broadband Technologies
21 Madbury Rd., Ste. 100, Durham, NH 03824
+1-603-674-2755 (m)



--
Brandon Lo
UNH InterOperability Laboratory
21 Madbury Rd, Suite 100, Durham, NH 03824
blo@iol.unh.edu
www.iol.unh.edu