Community CI Meeting Minutes - March 19, 2026

[Patrick Robb]

March 19, 2026
Attendees
   5. Patrick Robb
   6. Aaron Conole
   7. Adam Hassick
   8. Dean Marx
   9. Andrew Bailey
   10. Lincoln Lavoie
   11. Luca Vizzarro
Minutes
General Announcements
   * AI Code Review:
   * Patrick has created a PR which adds the jenkins scripting for driving
the AI code review scripting, depending on the “gate conditions” described
in the previous meeting minutes
   * Adam added as reviewer
   * Aaron to come into the UNH Lab soon to review the AI code review
process with the team.
   * Aaron will probably come in the latter portion of next week to
finalize our CI process / look over the scripting with us and then we can
launch the AI code reviews in production
   * We are swapping in Aaron’s review-start.md in place of Stephen’s
AGENTS.md because we are token constrained (need to spend less per patch)
   * In the long term there will be a CI usable AGENTS.md which is merged
into DPDK and we will just use this
   * RC2 was released this week
   * RC3 will be released next Tuesday
CI Status
AWS Lab
   * None, they are running normally
UNH-IOL Community Lab
   * Cryptodev testing:
   * The cryptodev series was merged in RC2, which modifies the
test_run.yaml “schema” so this is breaking. We are using new config file
which align with the new schema. A few next branches in DPDK have not
rebased off of main, which means that CI runs which apply to these branches
cannot run DTS. Patrick to check which branches are not rebased and ask the
maintainers to rebase on main.
   * Adding the new DTS testsuite to CI:
   * Andrew is staging adding the new testsuite to our CI runs. This will
be reporting to patchwork very soon for Intel QAT, then we will add other
cryptodevs.
   * AMD cryptodev: UNH Lab has a system with an AMD cryptodevice which is
included in the processor package, and has a DPDK poll mode driver. Andrew
to test adding this to our CI runs after he gets the Intel QAT online.
   * Rte_flow testsuite:
   * Merged to mainline, all devices have at least 1 failure, so we cannot
activate this series in CI
   * DTS failure ticketing:
   * We have been documenting some of the failures we see from DTS in the
ethdev component on Bugzilla along the way, but could be more intentional
about it. I.e. we should have a little downtime in the next couple of
weeks, it will be good if we can set aside time to do some testruns and
document the existing failures better.
   * Windows:
   * Server 2022 machine officially turned off - all CI runs moved to
server 2025
   * Getting a license for server 2025 from UNH
   * Bugzilla:
   * On the lab component, there are a bunch of old, unresolved tickets
that basically need to be updated or marked as resolved or will not do.
   * Once this is done, the lab component will be more usable again
Intel Lab
   * None
   * No news so far from Manju regarding their DTS usage - so gather that
is a work in progress
Github Actions Robot
   * None
Loongson Lab
   * None
DTS Improvements & Test Development
   * RC2
   * Merged our main initiatives per the roadmap
   * Patrick to email Maxime and notify him that we will merge the PVVP
virtio testcase at the start of the 26.07 window
   * RC3
   * Dean sent a bugfix for the logic which drives the overall testsuite
level outcome when there is a mix of PASS and SKIP results (PASS should
win).
   * Cryptodev:
   * Merged
   * Flow Offload:
   * Merged
   * https://bugs.dpdk.org/show_bug.cgi?id=1584
   * When we write dataclasses for a particular component in testpmd, it
makes sense to comprehensively support all valid fields allowed by testpmd,
not just the ones being used in our testsuites. This is what we have been
doing in most cases, but we should be doing it in all cases.
   * Patrick’s talk for DPDK Stockholm about integrating DTS in the
developer workflow (and demo-ing the rte flow testsuite) was accepted
yesterday.
   * Speed capabilities testsuite to be merged for RC3

Any other business
      * Next meeting is Mar 19, 2026
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mails.dpdk.org/archives/ci/attachments/20260402/ac50b1bc/attachment-0001.htm>