Community CI Meeting Minutes - May 30, 2024

Community CI Meeting Minutes - May 30, 2024

[Patrick Robb]

[-- Attachment #1: Type: text/plain, Size: 5189 bytes --]

May 30, 2024

#####################################################################
Attendees
1. Patrick Robb
2. Adam Hassick
3. Nicholas Pratte
4. Paul Szczepanek
5. Jeremy Spewock
6. Dean Marx
7. Aaron Conole
8. Juraj Linkeš

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

=====================================================================
General Announcements
* Server refresh at UNH. We have ordered the Intel and AMD servers. The ARM
server is not yet ordered but we are working with Supermicro.
* DPDK APAC Summit: Bangkok July 9-10, anyone who is able to attend, please
see here: https://sessionize.com/dpdksummitAPAC2024
* OvS/OvN Fall conference will be in person/hybrid, taking place is Europe.
* Dates are still being worked out for DPDK Summit in Montreal this Fall.
Stay tuned: https://www.dpdk.org/events/

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

---------------------------------------------------------------------
UNH-IOL Community Lab
* Targeting deploying SPDK compile testing in our CI by end of day.
* For projects downstream of DPDK, do we want to report warns instead of
fails in all cases? So this would apply to both SPDK and OvS.
   * Aaron agrees. We should not gate a patch because it causes an issue in
a downstream patch, but at the same time it should be flagged with a WARN
because it is likely that a patch causing issues in a downstream project
may cause other negative implications.
* PENDING results:
   * Small update needed on the pw server email parsing script
https://patchwork.dpdk.org/project/ci/patch/20240523215945.16468-1-probb@iol.unh.edu/
   * DTS Functional results now print a table in the report
   * Retests: It now includes in the email that results are a part of a
retest now
* Luca confirmed that we should use the full dts-format-patch.sh script
instead of just running mypy for changes coming into dts
* Ubuntu 24.04 environment has been added for compile and unit testing:
   * X86_64 is online right now, and we are also adding an arm64
environment next
* Depends-on:
   * Currently, Github Robot is the only lab with this supported.
   * IOL is targeting supporting this by mid July. This will involve
submitting Depend-on support to the pw server project, which their
maintainer has endorsed.
   * Currently on dpdk.org we post support for retests, but not for
depends-on.
      * Patrick Robbcan submit a little blurb noting how depends-on works,
how CI labs should treat it, and which labs currently support it

---------------------------------------------------------------------
Intel Lab
* CentOS 7 environment is now removed from compile testing matrix

---------------------------------------------------------------------
Github Actions
* Aaron spent some time in the past couple days fixing issues
post-migration.
   * There weren’t enough resources given to the GH Robot, so it ended up
crashing a few times. There may have been some missing reports on the DPDK
side. Retests are working if anyone needs new results.
* Cirrus CI support: Aaron plans to get back to this soon, but was
sidetracked in the previous month due to some customer needs
* Retests: Aims to implement the numbering scheme soon

---------------------------------------------------------------------
Loongarch Lab
* None

=====================================================================
DTS Improvements & Test Development
* Which patches are ready to be merged?
   * Mypy needs a V3 which some small adjustments, but I gather it will be
good to merge after that based on Juraj’s comments.
https://patchwork.dpdk.org/project/dpdk/patch/20240514113446.1948050-2-luca.vizzarro@arm.com/
   * Capabilities Checking:
      * Patrick ran this across Mellanox NIC and the behavior was as
expected for the scatter suite, will try other NICs today
         * What is the difference between these commands below, and which
is more appropriate for checking the device capability?
            * “show rxq info 0 0”
               * Prints result according to whether –enable-scatter
included as testpmd parameter
            * testpmd> show port (port_id) rx_offload capabilities
            * Questions for this can be directed to Ferruh, as he is a
maintainer for ethernet API
               * Thomas and Andrew Rybchenko also maintainers
      * Juraj needs to catch up on reviews right now, so there will be a
small delay before he can get back to this.
* Jumboframes: Luca requested that this be resubmitted, rebased off of the
testpmd parameters patch to validate that patch and make sure it is
compatible. But, we are not able to submit it in this fashion because we
already have a depends-on
* Testpmd parameters:
   * Has been validated by Nick in his jumboframes version which has not
been submitted, and he agreed that it makes the testsuite development
experience easier
   * New version coming pending most recent reviews

=====================================================================
Any other business
* Next Meeting: June 13, 2024

[-- Attachment #2: Type: text/html, Size: 5775 bytes --]