Community CI Meeting Minutes - March 30, 2023

Community CI Meeting Minutes - March 30, 2023

[Patrick Robb]

[-- Attachment #1: Type: text/plain, Size: 6048 bytes --]

March 30, 2023

#####################################################################
Attendees
1. Patrick Robb
2. Lincoln Lavoie
3. Aaron Conole
4. Lijuan Tu
5. Ali Alnubani
6. Adam Hassick
7. Wathsala Wathawana Vithanage
8. Honnappa Nagarahalli
9. Juraj Linkeš

#####################################################################
Agenda
1. General Announcements
2. CI Status
3. DTS Improvements & Test Development
4. Any other business

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

=====================================================================
General Announcements
* Effort improved by gov board: Final deadline for NIC hardware refresh
requests is 4/7 (gold members). Once that deadline is reached we will
proceed.
https://docs.google.com/spreadsheets/d/1nrXmPY4wi5vffTHExEkHXxKpnk5aGjjkYKGj2Gn3_nY/edit?usp=sharing
* Nathan sent out a survey about the DPDK Userspace Summit to determine
who/how many people will be attending:
https://docs.google.com/forms/d/1104swKV4-_nNT6GimkRBNVac1uAqX7o2P936bcGsgMc

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

---------------------------------------------------------------------
UNH-IOL Community Lab
* ABI Testing: On Tuesday a patch was merged to DPDK main which changes the
abidiff process for checking for ABI breakage (now comparing .so files
directly). Our process for ABI testing is to copy the ABI reference into
our container images in order to avoid compiling two times per ABI test.
So, we will have to rebuild our container images before bringing our ABI
testing back online.
   * Advance warning for such changes helps reduce downtime!
   * Need to account for LTS branches still using the old process
* Broadcom-25G performance testing on the Arm Gigabyte server is showing
greater variance than anticipated. We’ve tried isolating additional cores
for this testing, to no avail. This test coverage has been taken offline
for the time being.
   * It may be worth going to ARM/Broadcom for advice on device tuning
   * Lab can easily give VPN access to community members to remote into the
servers here if they wish to do any tuning themselves
* Aarch64 containerized unit testing: We are still debugging the behavior
seen from the eal_flags_file_prefix_autotest when running in a container,
with the help of ARM. In the meantime, we want to bring coverage for
aarch64 online for unit tests which are running without issue. The lab
successfully dry ran this coverage yesterday and will be bringing this to
production today for:
   * CentOS 9
   * Debian 11
   * Ubuntu 20.04
   * Fedora 36 (gcc)
   * Fedora 36 (clang)
* FIPS ACVP tool: We received some feedback on the patch submission about
some potentially unneeded dependencies for this app. We’re dry running a
FIPS app pipeline internally with the suggested removals and will submit a
v7 patch series according to the results of these dry runs.

---------------------------------------------------------------------
Intel Lab
* Aaron: Greater than normal amount of failures seen recently?
   * Lijuan: This is related to their CI infrastructure. Normally baseline
is sent out to the community, but with the new containerized test bed,
utilizing new baselines was overlooked. This has now been addressed.

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

---------------------------------------------------------------------
Github Actions
* Machines are being moved out of 1 lab and moved to a different physical
lab space. Not sure when this is happening. For a period of time the GHA
robot will be down - Aaron will update with ETA later. Migration should be
fairly simple and seamless.
   * Best case: 1 day of downtime (estimate)

=====================================================================
DTS Improvements & Test Development
* ‘Hello World’ patch merged in 23.03
* Roadmap for 23.07 release - 15th April
   * Work complete on tooling for developing python documentation - Google
docstring format selected. Patch sent to the community.
      * Needs reviews on the meson integration.
      * One module re-written to use the Google docstring.
      * Uses the same Spinx config file as the one used for DPDK
documentation. This should not affect DPDK documentation and is isolated
for DTS.
      * Separating the DPDK doc and DTS doc is required. Added the
‘enable-dts’ option, but it does not seem to work.
   * Non-automated documentation such as coding guidelines need to be added
   * Fabric integration is completed. Doc changes need to happen. This will
contain the passwordless sudo, removing the use of scp.
   * Smoke tests
      * Patrick will talk to Jeremy to follow up on the plan over the
email. As a suggestion, non-TG smoke tests would be an option for this
release.
   * TG abstraction could be an option. Juraj has looked at it and might be
able to do this. Try to create a RFC if possible.
   * Minor changes
      * Using a git commit ID instead of tarball.
      * Updates to the linter tools.
      * Logging changes - each testsuite logging into a separate file.
   * Patches that were not merged - docker file
* Jeremy will not be able to attend the calls. Typically works from noon to
6pm Wendesdays.

=====================================================================
Any other business
* Next meeting is April 13th
* Labs will be discussed in Gov board meeting in April
   * Information desired about power outage
      * UNH lab has no way around power outages - a generator backed
datacenter is not financially viable. The best we can do is backfill test
results when power is restored
      * This was the most serious power outage seen at the lab in the past
few years

-- 

Patrick Robb

Technical Service Manager

UNH InterOperability Laboratory

21 Madbury Rd, Suite 100, Durham, NH 03824

www.iol.unh.edu

[-- Attachment #2: Type: text/html, Size: 8642 bytes --]