Fix #14711 cherry pick from #15850#15946
Conversation
What is the motivation for this PR? sonic-net#11169 How did you do it? Add a new test case to cover "pfcwd show stats" cmd Trigger PFC storm with action drop or forward send_tx_egress and send_rx_ingress execute CMD 'show pfcwd stat' and ensure the output is expected How did you verify/test it? https://elastictest.org/scheduler/testplan/66f268ae4216a91fd43d97b6 https://elastictest.org/scheduler/testplan/66f268880369ccd340b3ea60
Signed-off-by: Austin Pham <austinpham@microsoft.com>
|
@auspham, @yejianquan May I know why this PR changed the code for the Mellanox part? The pfcwd_function test has been not stable on Mellanox platform since last week. I'm still doing some investigation. |
|
After some investigation, I think the change on the Mellanox part caused some issue. sonic-mgmt/tests/pfcwd/test_pfcwd_function.py Lines 739 to 741 in 15bf903 I will raise a PR to revert the Mellanox related change. Please comment if you have any concern |
Hi @bingwang-ms, this is to cherry-pick #14711 since the branch has some conflict that cannot auto-resolve. The purpose is to merge in #15841. |
|
@auspham , @yejianquan Can you please help review PR? Please help check the change doesn't cause regression on chassis. I didn't get a testbed to verify. |
Description of PR Summary: This PR is to fix test issue caused by PR #15946 The code below is to ensure there is PFCWD triggered on new queues before checking syslog. sonic-mgmt/tests/pfcwd/test_pfcwd_function.py Lines 739 to 741 in 15bf903 pytest_assert(wait_until(PFC_STORM_TIMEOUT, 2, 0, lambda: check_pfc_storm_state(dut, port, self.storm_hndle.pfc_queue_idx) != pfcwd_stats_before_test), # noqa: E501, E128 "PFC storm state did not change as expected") # noqa: E127 The test is flaky after change in #15946 because the wait time is not enough for PFCWD to be triggered on new queues. Approach What is the motivation for this PR? This PR is to fix test issue caused by PR #15946 How did you do it? Change the logic back while maintain compatibility with chassis. How did you verify/test it? The change is verified on a Mellanox platform. collected 5 items pfcwd/test_pfcwd_function.py::TestPfcwdFunc::test_pfcwd_actions[str2-msn4600c-acs-04] ^H ^H ^H ^HPASSED [ 20%] pfcwd/test_pfcwd_function.py::TestPfcwdFunc::test_pfcwd_multi_port[str2-msn4600c-acs-04] ^H ^H ^HPASSED [ 40%] pfcwd/test_pfcwd_function.py::TestPfcwdFunc::test_pfcwd_mmu_change[str2-msn4600c-acs-04] ^H ^HPASSED [ 60%] pfcwd/test_pfcwd_function.py::TestPfcwdFunc::test_pfcwd_port_toggle[str2-msn4600c-acs-04] ^H ^H ^HPASSED [ 80%] pfcwd/test_pfcwd_function.py::TestPfcwdFunc::test_pfcwd_no_traffic[str2-msn4600c-acs-04] SKIPPED (This test is applicable only for cisco-8000 / Pfcwd tests skipped on m0/mx testbed.) [100%] Any platform specific information? Mellanox platform specific. co-authorized by: jianquanye@microsoft.com
4 participants
Description of PR
Summary:
Fixes # (issue) #15850
Cherry-pick #14711 to 202405
Failure is introduced from https://github.com/sonic-net/sonic-mgmt/pull/14580/files
Currently blocking #15841
Type of change
Back port request
Approach
What is the motivation for this PR?
#11169
How did you do it?
Add a new test case to cover "pfcwd show stats" cmd
Trigger PFC storm with action drop or forward
send_tx_egress and send_rx_ingress
execute CMD 'show pfcwd stat' and ensure the output is expected
How did you verify/test it?
https://elastictest.org/scheduler/testplan/66f268ae4216a91fd43d97b6
https://elastictest.org/scheduler/testplan/66f268880369ccd340b3ea60
Any platform specific information?
Supported testbed topology if it's a new test case?
Documentation