March 2, 2023 ##################################################################### Attendees 1. Patrick Robb 2. Honnappa Nagarahalli 3. Lincoln Lavoie 4. Juraj Linkeš 5. Aaron Conole 6. Adam Hassick ##################################################################### Agenda 1. General Announcements 2. CI Status 3. DTS Improvements & Test Development 4. Any other business ##################################################################### Minutes ===================================================================== General Announcements * None ===================================================================== CI Status --------------------------------------------------------------------- UNH-IOL Community Lab * We are done expanding test vector coverage on our FIPS sample app CI testing. We are monitoring the published results and will submit a patch to upstream our acvp tool/configs/documentation to the dpdk-ci repo next week. This will allow others to more easily test crypto implementations in a CI environment if they are interested. * We do see active FIPS development on the next-crypto branch, and will expand our test vector coverage as the sample app evolves. * We are now using the most up to date “guess git tree” script as a part of our new containerized apply patchset process. * Compiler cache is re-enabled for our CI, which will help reduce the time to delivering test results. * ARM-Gigabyte began exhibiting booting issues two weeks ago and was taken offline. We had to remove the RAM and PCI cards to get the BMC and BIOS firmware upgraded to the latest releases (which is completed now), and then reseated all cards / RAN, and starting yesterday it has been reporting test results again. * We have added Broadcom-25g testing to this test bed, which is working internally. I will publish a performance baseline for this nic/server combination and bring it online for public reporting today * Unit_test_mbuf failures seen on the ARM Altra system are not appearing on the Gigabyte system, indicating it’s not a universal ARM issue. ARM has also tested internally and are not yet replicating the problem - UNH will continue to look into this. * UNH has changed to submitting plain text emails, which has improved formatting but not resolved the DMARC policy issue, which may require shortening the subject line or other refactoring. * We can cap the subject to 70 characters to resolve this issue * We are dry running our new multi-arch testing pipelines and addressing some remaining bugs. This should be online next week. Once that’s done, we’ll also send up a next version of the patch series to dpdk-ci for the docker manifest based builds (i.e. multi-arch support). * Gold members are being polled by Nathan regarding their preferences for a hardware refresh. * Can/should GPUs be a part of the hardware refresh? First focus is on NICs. * How would we handle new NICs which cannot run on our current servers due to pcie slots? --------------------------------------------------------------------- Intel Lab * We have enabled a container testbed to cover more functional test cases. And reports are sending to patchwork. --------------------------------------------------------------------- Loongarch Lab * None --------------------------------------------------------------------- Github Actions * One change for patchwork ci script to support events api polling series completed instead of polling the patch series themselves. This is going in today or tomorrow - was reviewed this morning. * Will next integrate with the newest guess git tree script. They will be pulling directly from the most up to date dpdk-ci public repo. ===================================================================== DTS Improvements & Test Development Patch discussions * Hello World patch 1. Bruce has reviewed the patch. Wathsala has tested on Arm platforms. 2. Wathsala’s comments were on the roadmap for the next release 3. Waiting on Lijuan’s review and testing at Intel lab DTS Developer documentation * Docstring format * Google * Best suited for our needs * NumPy * Better suited for projects with very long docstring (a lot of explaining needs to be done) * Sphinx * Basic format, more cluttered than the two above * Other non-Sphinx format I found were either legacy or Java-based * Tools * DPDK uses Doxygen and Sphinx, the above tools use Sphinx so we can reuse that. * We’ll go with the Google format, reusing DPDK Sphinx integration. AI Juraj to create a patch * It is good to bring up the discussion with the community about the format if DTS differs from what DPDK uses. Continuing Owen’s work * UNH had to go back to hiring as the candidate did not join * Jeremy is ramping up to look at smoke tests for DTS 23.07 Roadmap * This list will evolve. Following is a list of things we want to do for sure. * Incorporating Fabric. Documentation on how to setup password-less sudo on the SUT * DTS Developer documentation - agreeing on doc-strings to use * Node discovery feature - Do design to identify the needs of DUT and tester? * Non-TG blocked DTS smoke tests? * Unit tests for DTS? Review Action Items * Juraj - Developer documentation patch - Reformat docstrings, integrate with Sphinx * Honnappa - Follow up with Tech Board and Thomas on DTS maintainer in DPDK - Done ===================================================================== Any other business * Next meeting is March 16th -- Patrick Robb Technical Service Manager UNH InterOperability Laboratory 21 Madbury Rd, Suite 100, Durham, NH 03824 www.iol.unh.edu