Community CI Meeting Minutes - March 21, 2024

[Patrick Robb <probb@iol.unh.edu>]

March 21, 2024

#####################################################################
Attendees
1. Patrick Robb
2. Aaron Conole
3. Nicholas Pratte
4. Jeremy Spewock
5. Cody Cheng

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

=====================================================================
General Announcements
* Daylight savings time has hit North America, and will also happen in
Europe between this meeting and the next one. Should we adjust?
   * Going to push this decision back two weeks, and decide at the
next CI Meeting
* IPSEC-MB requirement increase:
   * This has been postponed to a future release
   * UNH did confirm that the final tag submitted by ARM for their
repo (tag: SECLIB-IPSEC-2024.03.12) is compatible with the proposed
changes from Intel in that DPDK now compiles, with tests passing.
   * V1.5 will ship with Ubuntu 24.04. Fedora already has it. Once it
is widely available with latest versions of standard distros, this
requirement will likely be reconsidered.
* Server Refresh:
   * Aaron and Patrick presented this idea to Governing Board. It is
still under consideration.
      * There was a followup provided showing which specific servers
will have to be retired, which NICs will be migrated between servers,
etc.
         * We can still use these systems as dev systems and/or SW
based testing.
      * We are currently working with ARM to get a quote for a Grace
system so we can follow up with GB for that too
         * Need to figure out what the lead time is for this server to ship
* We have an agreed upon process and format for v2 of the Email Based
Retest Framework. Zhoumin at Loongson is also working on the lab
specific implementation needed in order for him to start using the
get_reruns.py script and support retests.
   * https://inbox.dpdk.org/ci/de103a95-6f57-df25-0ab5-344dfd8a391a@loongson.cn/T/#t
   * Final conclusion is like:
      * A) If retest is requested without rebase key, then retest
"original" dpdk artifact (either by re-using the existing tarball (unh
lab) or tracking the commit from submit time and re-applying onto dpdk
at that commit (loongson)).
      * B) If rebase key is included, apply to tip of the indicated
branch. If, because the branch has changed, the patch no longer
applies, then we can report an apply failure. Then, submitter has to
refactor their patch and resubmit.
      * In either case, report the new results with an updated test
result in the email (i.e. report "_Testing PASS RETEST #1" instead of
"_Testing PASS" in the email body).
* Depends-on support: The Patchwork maintainer accepted the plan for
adding depend-on support to PW server. But, someone needs to write the
code and submit it to the Patchwork project. Adam has indicated his
willingness, but we want to check with the DPDK Community that it is a
good value from the community’s perspective before we spend a lot of
community lab resources on this. His email didn’t get a response but
we’ll re-ping the thread after the 24.03 release when people are a
little less busy.
   * https://inbox.dpdk.org/ci/CAC-YWqhFpn5zrZsJ_R6sUehH1FzSWpuRFpgQ-CRaG5p8NrHAbQ@mail.gmail.com/T/#u
   * https://github.com/getpatchwork/patchwork/issues/583 and
https://github.com/getpatchwork/git-pw/issues/71
* There will (probably) be a DPDK event in Montreal this Sept 9-11.
   * UNH will try to send as many people as possible, since we are so close

=====================================================================
CI Status

---------------------------------------------------------------------
UNH-IOL Community Lab
* Hardware Refresh:
   * QAT 8970 is enabled in CI, running the crypto_perf_cryptodev_perf
DTS testsuite.
* Windows updates:
   * Unit tests, Clang compile Mingw64 compile are all added to the
Windows 2022 machine
* Cody got confirmation from the 23.11 LTS maintainer that the staging
branch is now triggering when he pushes.
* OvS DPDK testing:
   * What are we actually testing? Regressions coming in from DPDK
patches, or latest state of x OVS branch?
      * The former
   * Performance testing requires some baremetal testbed topology,
where we would send some traffic, run OvS on the DUT, pass packets to
OvS in a VM, TestPMD forwards packets (or something).
      * Need to verify that the Community Lab is the correct setting
for this testing, that we are able to provide meaningful and stable
results. We just need to define more specifically what the test would
look like, and then we will know.

---------------------------------------------------------------------
Intel Lab
* None

---------------------------------------------------------------------
Github Actions
* Yesterday there was an outage at Ubuntu for their main package
archives for a few hours, so the Robot was failing due to this
dependency. We will have to put in retests.
* There was another failure in the last two weeks due to update of the
Ubuntu container image.
   * https://github.com/actions/runner-images/issues/9491
   * Due to kernel update for increased ASLR entropy
   * LLVM Asan using a region that it shouldn’t, causing a crash
   * Workaround pushed to DPDK archives, and github has updated their
base images to include the workaround as well
Loongarch Lab
* Zhoumin is working on the Loongson lab specific implementation of
the retest framework (will still leverage the dpdk-ci get_reruns.py
script). He states he will be able to support retests “as-is” and also
rebase on tip of branch options.
DTS Improvements & Test Development
* Roadmap is ready
* Ferruh was asking about DTS coverage for flow rules. It is outside
the scope of the work in the lab SOW and not in the 24.07 roadmap, but
we will pick it up if we are able. As writing a basic suite to
validate some flow rules would just require starting testpmd, sending
some commands in the testpmd shell setting flow rules, sending
packets, validating that the rules were followed - because DTS
supports all this, writing a testsuite should be possible currently.
* The Scatter suite was merged
* Nick has run his patch fixing the Hugepages allocation bug in DTS -
will submit this once the 24.07 cycle begins. In the interim he is
moving on to the other config refactoring related tickets on bugzilla
assigned to him.

=====================================================================
Any other business
* Next Meeting: April 4, 2024