[dpdk-ci] June 17 Community CI Meeting Minutes
Lincoln Lavoie
lylavoie at iol.unh.edu
Fri Jun 18 03:55:17 CEST 2021
Hi Zhaoyan,
Owen from our team has been leading the efforts on getting ASAN running on
(under, with) DPDK. He's got a couple of patches submitted fixing some
smaller issues he's seen. I think David Marchand also had some patches
started to fix some of the issues as well.
Cheers,
Lincoln
On Thu, Jun 17, 2021 at 8:39 PM Chen, Zhaoyan <zhaoyan.chen at intel.com>
wrote:
> Hi, Lincoln,
>
>
>
> For * ASAN deployment, what kind of tests are you going to deploy?
>
>
>
> We have a patch for enable ASAN for DPDK. I don’t know if is it helpful..
>
>
> http://patches.dpdk.org/project/dpdk/patch/20210615081205.101071-1-zhihongx.peng@intel.com/
>
>
>
> Thanks
>
>
>
> /Zhaoyan
>
>
>
> *From: *ci <ci-bounces at dpdk.org> on behalf of Lincoln Lavoie <
> lylavoie at iol.unh.edu>
> *Date: *Thursday, June 17, 2021 at 23:48
> *To: *"ci at dpdk.org" <ci at dpdk.org>
> *Subject: *[dpdk-ci] June 17 Community CI Meeting Minutes
>
>
>
> June 17, 2021
>
> #####################################################################
> Attendees
>
> 1. Lincoln Lavoie
> 2. Brandon Lo
> 3. Owen Hilyard
> 4. Aaron Conole
> 5. Ali Alnubani
> 6. Juraj Linkeš
> 7. Lijuan Tu
> 8. Ashley Weltz
> 9. Thomas Monjalon
> 10. Michael Santana
> 11. David Marchand
>
> #####################################################################
> Agenda
>
> 1. Performance Metric Change to Percentage
> 2. Coverity Testing
> 3. CI Lab Status
> 4. Test Development
> 5. Any Other Business
>
> #####################################################################
> Minutes
>
> =====================================================================
> Performance Metric Change to Percentage
>
> * Patch series is still pending merge:
> https://patchwork.dpdk.org/project/dts/list/?series=15990
> * Blocked by: https://bugs.dpdk.org/show_bug.cgi?id=670
> * Blocks any upgrade to the latest DTS versions, which depend on the
> latest TRex.
> * A work around could be to skip testing specific frame sizes that cause
> that issue. Ali will confirm internally and reply back to the lab on this
> way forward.
>
> =====================================================================
> Coverity Testing
>
> * https://bugs.dpdk.org/show_bug.cgi?id=719
> * Need to investigate if the lab can set up a way to desktop analysis,
> which would run local to the systems. Request is to run the test per
> patch, to check for additional issues being introduced by the patch. Only
> reports new defects.
> *
> https://community.synopsys.com/s/topic/0TO34000000LJdaGAG/coverity-desktop-analysis
> I think this is not an article (just Q&A), you might have to check the
> documentation in your local installation
> * Need to verify the requirements for running the tool, and if it required
> a paid account, etc.
>
> =====================================================================
> CI Lab Status
>
> ---------------------------------------------------------------------
> UNH-IOL Community Lab
>
> * Currently looking into 40G Intel machine issues.
> * Stats checks disabled due to the requirement for VM images for the
> test suite in newer DTS versions
> * MTU update test suite disabled due to issue of unknown origin causing
> packets to appear as received without appearing as sent by the tester.
> * UNH to look at enabling an SSH jump host server, to work around VPN
> issues.
> * Dynamic Config DTS test case is still disabled
> * https://bugs.dpdk.org/show_bug.cgi?id=639
> * Need to confirm the patch status within DTS, looks like it was merged
> 8-days ago.
> * Mellanox testing disabled
> * DTS seems to be sending packets out the wrong ports, flipping TX/RX
> ports in code seems to fix this. Root cause is unknown. Can not upgrade
> DTS / TRex because of the blocking issue listed above.
> * Unit testing failures
> * Still seeing random failures on the Func_reentrancy_autotest unit
> test. 3 of the last 5 unit test failures are this test. Failures do not
> seem related to OS, Virtual or Bare-metal, or architecture.
> * Aaron is going to create a bug for this, and will look into removing
> the test case from the suite for now.
> * ASAN deployment
> * Various memory leak fixes in patchworks
> * rte_pci_scan
> *
> http://patches.dpdk.org/project/dpdk/patch/20210616065257.16044-1-david.marchand@redhat.com/
> * ribv6
> *
> http://patches.dpdk.org/project/dpdk/patch/20210616181833.356159-1-ohilyard@iol.unh.edu/
> * cmdline autotest
> *
> http://patches.dpdk.org/project/dpdk/patch/20210616180724.355217-1-ohilyard@iol.unh.edu/
> * test_eal_flags
> *
> http://patches.dpdk.org/project/dpdk/patch/20210616162453.349614-1-ohilyard@iol.unh.edu/
> * flow_classify
> *
> http://patches.dpdk.org/project/dpdk/patch/20210616195724.366103-1-ohilyard@iol.unh.edu/
> * Still more leaks in unit tests, which prevent deployment.
> * Periodic Testing
> * Run daily, but only if the git HEAD has moved.
> * Branches (pointing to the default of the repos):
> * dpdk main branch
> * dpdk-next-net
> * dpdk-next-virtio
> * dpdk-next-eventdev
> * dpdk-next-crypto
> * LTS current and previous (i.e. DPDK release plan) (e.g. 20.11 and
> 19.11 currently)
> * Test cases:
> * mtu_update
> * scatter
> * unit_tests_mbuf
> * stats_checks
> * dynamic_config
> * Planned, but disabled due to
> https://bugs.dpdk.org/show_bug.cgi?id=639
> * mac_filter
> * link_status_interrupt
> * nic_single_core_perf
> * Should the lab run ABI/Compile/Unit testing periodically?
> * Daily, same triggers as described above
> * When and where should the lab send an email?
> * Every run (both pass & fail)
> * On failure
> * Something else?
> * To maintainers, devs mailing list, somewhere else?
> * Send the reports to the test-reports mailing list. In the future
> we will need to align the report formatting between labs. CC the
> maintainers for what was tested..
> * How often should these be run?
> * Daily if there are any changes (Current)
> * rte flow
> * Will be deployed alongside periodic testing, to provide a way to
> test systems while failures are addressed by the maintainers
> * Clang Cross-compile Testing
> * Failure to build. Related to event/cnxk/cnxk_tim_worker.
> * Bug submitted to https://bugs.dpdk.org/show_bug.cgi?id=740
> * Production pipeline is set up to enable this test once this bug is
> resolved
> * FreeBSD 13
> * Now doing compile and ABI testing
> * Unit testing blocked by contigmem-related issue
> * Clock skew issue resolved
> * Patch applied to IOL infrastructure and merged into DTS
> * ABI False failures
> * Last week (June 4), a couple of false failures were caused by
> compiler updates (i.e. updating container images after the ABI reference
> was generated). The lab saves off the reference as an optimization. The
> lab systems were updated to regenerate the ABI references if container
> images are updated.
> * aarch64 SVE testing
> * With Ccache enabled, builds take about 5 to 7 minutes within the
> emulated environment. This will keep the compile and binarys in the same
> environment (i.e. within a single container).
>
> ---------------------------------------------------------------------
> Intel Lab
>
> * Patchwork Series https://bugs.dpdk.org/show_bug.cgi?id=718
> * It’s on-going, and we have found the root cause. The solution is a
> little complicated in CI script, and needs more time to fix them.
> * Some of the Intel test bed is off-line (for valid reasons, i.e.
> debugging or development), this can block the reporting. Some workaround
> have been made, will disable the test-bed if there is a need to offline it.
> And if it is offline unexpectedly, CI will notify CI maintainers.
> * Test reports mailing list was created for DTS, but patchwork needs to be
> set up to monitor this list. Lijuan will send an example email report to
> Ali.
> * http://mails.dpdk.org/archives/dts-test-report/2021-June/000005.html
> * http://mails.dpdk.org/archives/dts-test-report/2021-June/subject.html
> * Also need to add the email list onto the webpage for
> https://www.dpdk.org/contribute/#mailing-lists
>
> ---------------------------------------------------------------------
> Github Actions / OBS
>
> * Github no changes.
> * OBS no changes.
>
> =====================================================================
> CI / Test Development
>
> * DTS Improvement
> * Discussions are progressing. Work is being tracked here:
> https://docs.google.com/document/d/1c5S0_mZzFvzZfYkqyORLT2-qNvUb-fBdjA6DGusy4yM/edit
> * Next Meeting: TBD (working around vacation schedules)
> * Developer Retesting Request
> * https://mails.dpdk.org/archives/ci/2021-May/001189.html
> * Seeking feedback on the “command syntax.” Still need to implement
> the tracking / mechanics for things like patches and message ID tracking.
> * Community Lab Objectives
> * Q1 Update to the techboard:
> https://docs.google.com/document/d/1v0VKtZdsMXg35WNDawdsnqj5J4Xl9Egu_4180ukKD2o
> * Status tracking is also being partly carried out in Bugzilla. Epics
> have been created for each major objective. This will span 2021, do not
> expect them to be closed quickly. Updates posted to each objective will
> track story completion, etc. This is a compromise of how to easily track
> the long term objectives.
> * Open Requests (outside of planned objectives)
> * Add meson syntax check testing (
> https://bugs.dpdk.org/show_bug.cgi?id=696)
> * Will leave this open for now for folks to give feedback on.
>
> =====================================================================
> Any Other Business
>
> * Next CI Meeting: June 30, 1pm UTC
>
>
>
> --
>
> *Lincoln Lavoie*
>
> Principal Engineer, Broadband Technologies
>
> 21 Madbury Rd., Ste. 100, Durham, NH 03824
>
> lylavoie at iol.unh.edu
>
> https://www.iol.unh.edu
>
> +1-603-674-2755 (m)
>
> [image: Image removed by sender.] <https://www.iol.unh.edu/>
>
--
*Lincoln Lavoie*
Principal Engineer, Broadband Technologies
21 Madbury Rd., Ste. 100, Durham, NH 03824
lylavoie at iol.unh.edu
https://www.iol.unh.edu
+1-603-674-2755 (m)
<https://www.iol.unh.edu>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mails.dpdk.org/archives/ci/attachments/20210617/0e7c2bdd/attachment-0001.htm>
More information about the ci
mailing list