Update qos sai due to no lossless pg for some platforms #16440

JibinBao · 2025-01-10T03:53:58Z

Description of PR

For spc4 and above, there is only the lossy buffer, so the buffer for the lossless buffer will be taken by the lossy buffer. If the packet size is too small, the packet number sent to occupy the shared buffer will increase a lot, which will lead to the descriptor being exhausted, so update testQosSaiPgSharedWatermark, testQosSaiQSharedWatermark, and testQosSaiLossyQueue accordingly.
Remove the test config of scheduler.block_data_plane, otherwise it might raise yang validation error when do config reload
When there is no lossless buffer, return a dump buffer lossless pg profile, and skip tests related to lossless buffer case dynamically
Skip fixture reaseAllports for mellanox device, because after qos test is finished, the teardown will do config reload, it will restore the config of ports, we don't need this fixture before running tests. Also it can save 2 minutes
list the relevant Prs:
[Mellanox] Update buffer calculations for Mellanox-SN5600-C224O8 SKU sonic-buildimage#20992
[Mellanox] Add x86_64-nvidia_sn5610n-r0 new platform and SKUs sonic-buildimage#21056
[Mellanox] Update Mellanox-SN5600-C256S1 buffer calculations sonic-buildimage#20991
[Mellanox] Add x86_64-nvidia_sn5610n-r0 new platform and SKUs sonic-buildimage#21056
[Mellanox] Update Mellanox-SN5600-C256S1, Mellanox-SN5600-C224O8 buffers and DSCP mapping sonic-buildimage#21427
[Mellanox] Add x86_64-nvidia_sn5610n-r0 new platform and SKUs sonic-buildimage#21056
[Mellanox] Update DSCP mapping for SN5600, SN5610 SKUs sonic-buildimage#21762

Summary:
Fixes # (issue)

Type of change

Back port request

Approach

What is the motivation for this PR?

update the qos sai test for no pg lossless buffer platform

How did you do it?

update for lossy case and skip test relatd to pg buffer lossless

How did you verify/test it?

Run qos sai test on platform without pg lossless buffer plaform

Any platform specific information?

sn5600 and sn5610

Supported testbed topology if it's a new test case?

Documentation

mssonicbld · 2025-01-10T03:54:02Z

/azp run

azure-pipelines · 2025-01-10T03:54:13Z

Azure Pipelines successfully started running 1 pipeline(s).

JibinBao · 2025-01-10T06:49:48Z

/azpw run Azure.sonic-mgmt

mssonicbld · 2025-01-10T06:49:51Z

/AzurePipelines run Azure.sonic-mgmt

azure-pipelines · 2025-01-10T06:50:04Z

Azure Pipelines successfully started running 1 pipeline(s).

mssonicbld · 2025-02-19T07:51:50Z

/azp run

azure-pipelines · 2025-02-19T07:51:56Z

Pull request contains merge conflicts.

mssonicbld · 2025-02-19T08:14:22Z

/azp run

azure-pipelines · 2025-02-19T08:14:36Z

Azure Pipelines successfully started running 1 pipeline(s).

#### Why I did it To have DSCP mapping updated to Mellanox SN5600, SN5610N SKUs #### How I did it Update buffers_defaults_objects.j2 and qos.json.j2 according to new DSCP mapping #### How to verify it Check SDK dumps to make sure values are correct Also, run sonic-mgmt test. relevant changes can be found here - sonic-net/sonic-mgmt#16440  #### Which release branch to backport (provide reason below if selected)  - [ ] 201811 - [ ] 201911 - [ ] 202006 - [ ] 202012 - [ ] 202106 - [ ] 202111 - [ ] 202205 - [ ] 202211 - [ ] 202305 #### Tested branch (Please provide the tested image version)  - [ ]  - [ ]  #### Description for the changelog   #### Link to config_db schema for YANG module changes  #### A picture of a cute animal (not mandatory but encouraged)

mssonicbld · 2025-02-21T02:21:12Z

/azp run

azure-pipelines · 2025-02-21T02:21:23Z

Azure Pipelines successfully started running 1 pipeline(s).

bingwang-ms · 2025-02-26T19:09:32Z

@kperumalbfn , @XuChen-MSFT Can you please help review?

kperumalbfn · 2025-02-28T17:04:49Z

@JibinBao can we combine this with #17004 (comment)? Both the PRs are related and we can use the same logic to skip the lossless tests on SKUs with only lossy configs.

JibinBao · 2025-03-05T03:46:24Z

@JibinBao can we combine this with #17004 (comment)? Both the PRs are related and we can use the same logic to skip the lossless tests on SKUs with only lossy configs.

hi @kperumalbfn ,

We have one same logic:
https://github.com/sonic-net/sonic-mgmt/pull/16440/files#diff-fddd5bf26ca50aa4d3504424556801116946ed4d39d1824b343e39ebcc01d4ffR2132

profile_content = dut_asic.run_redis_cmd(
argv=["redis-cli", "-n", 0, "keys", f'BUFFER_PG_TABLE:{srcport}:*-4'])
if not profile_content:
is_lossy_queue_only = True
logger.info(f"{srcport} has only lossy queue")

Can we create a function with the above code and put it tests/qos/conftest.py

def detect_lossy_only_pool(dut_asic):
is_lossy_queue_only = False
profile_content = dut_asic.run_redis_cmd(
argv=["redis-cli", "-n", 0, "keys", f'BUFFER_PG_TABLE:{srcport}:*-4'])
if not profile_content:
is_lossy_queue_only = True
logger.info(f"{srcport} has only lossy queue")
return is_lossy_queue_only

1. For spc4, there is only lossy buffer, so the buffer for lossless buffer will be taken by lossy buffer, if the packet size is too small, the packet number sent to occupy the shared buffer will increase a lot, so it will lead that descriptor will be exhausted, so update testQosSaiPgSharedWatermark, testQosSaiQSharedWatermark and testQosSaiLossyQueue accordingly. 2. Remove the test config of scheduler.block_data_plane, otherwise it might rise yang validation error when do config reload 3. list the relevant Prs: sonic-net/sonic-buildimage#20992 sonic-net/sonic-buildimage#21056 sonic-net/sonic-buildimage#20991

1. When there is no lossless buffer, return a dump buffer lossless pg profile 2. Skip tests related to lossless buffer case dynamically

1. Update qos sai lossy tests because all lossless queues has changed as lossy queues 2. Skip fixture reaseAllports for mellanox device, because after qos test is finished, the teardown will do config reload, it will restore the config of ports, we don't need this fixture before running tests. Also it can save 2 minutes 3. Relevant PRs: sonic-net/sonic-buildimage#21427, sonic-net/sonic-buildimage#21056

…t/sonic-buildimage#21762

mssonicbld · 2025-03-05T08:48:32Z

/azp run

azure-pipelines · 2025-03-05T08:48:43Z

Azure Pipelines successfully started running 1 pipeline(s).

JibinBao · 2025-03-17T01:40:36Z

Hi @bingwang-ms, @kperumalbfn
Can you review it again?

bingwang-ms · 2025-03-26T16:28:46Z

@kperumalbfn Can you please help review?

kperumalbfn · 2025-03-26T20:24:03Z

tests/qos/qos_sai_base.py

        if srcport in dualtor_ports_for_duts:
            queues = "0-1"
        else:
-            queues = "0-2"
+            if isMellanoxDevice(duthost):
+                profile_content = dut_asic.run_redis_cmd(


@JibinBao Could we do similar to cable_length "0m" check instead of BUFFER_PG:4??

@kperumalbfn ,
Done. Can you please review it again?

Thanks @JibinBao

mssonicbld · 2025-03-31T09:44:06Z

/azp run

azure-pipelines · 2025-03-31T09:44:18Z

Azure Pipelines successfully started running 1 pipeline(s).

…6440) 1. For spc4 and above, there is only the lossy buffer, so the buffer for the lossless buffer will be taken by the lossy buffer. If the packet size is too small, the packet number sent to occupy the shared buffer will increase a lot, which will lead to the descriptor being exhausted, so update testQosSaiPgSharedWatermark, testQosSaiQSharedWatermark, and testQosSaiLossyQueue accordingly. 2. Remove the test config of scheduler.block_data_plane, otherwise it might raise yang validation error when do config reload 3. When there is no lossless buffer, return a dump buffer lossless pg profile, and skip tests related to lossless buffer case dynamically 4. Skip fixture reaseAllports for mellanox device, because after qos test is finished, the teardown will do config reload, it will restore the config of ports, we don't need this fixture before running tests. Also it can save 2 minutes 5. list the relevant Prs: [Mellanox] Update buffer calculations for Mellanox-SN5600-C224O8 SKU sonic-buildimage#20992 [Mellanox] Add x86_64-nvidia_sn5610n-r0 new platform and SKUs sonic-buildimage#21056 [Mellanox] Update Mellanox-SN5600-C256S1 buffer calculations sonic-buildimage#20991 [Mellanox] Add x86_64-nvidia_sn5610n-r0 new platform and SKUs sonic-buildimage#21056 [Mellanox] Update Mellanox-SN5600-C256S1, Mellanox-SN5600-C224O8 buffers and DSCP mapping sonic-buildimage#21427 [Mellanox] Add x86_64-nvidia_sn5610n-r0 new platform and SKUs sonic-buildimage#21056 [Mellanox] Update DSCP mapping for SN5600, SN5610 SKUs sonic-buildimage#21762

r12f · 2025-04-07T21:33:18Z

hi @JibinBao , do you mind to help create a manual merge PR to 202412? there is a merge conflict for this PR.

r12f · 2025-04-15T20:12:27Z

hi @JibinBao , do you mind to help create a manual merge PR to 202412? there is a merge conflict for this PR.

JibinBao · 2025-04-16T02:06:46Z

hi @JibinBao , do you mind to help create a manual merge PR to 202412? there is a merge conflict for this PR.

Hi @r12f , I don't find the 202412 branch. Do you mean msft_202412?

r12f · 2025-04-18T18:41:09Z

hi Jibin, yes. the repo is here: https://github.com/Azure/sonic-mgmt.msft

JibinBao · 2025-04-25T10:24:01Z

hi Jibin, yes. the repo is here: https://github.com/Azure/sonic-mgmt.msft

Hi @r12f Please review Azure/sonic-mgmt.msft#225

r12f · 2025-04-28T02:04:47Z

thanks Jibin. It is merged now.

JibinBao changed the title ~~Update qos sai due to no lossless pg for some platform~~ Update qos sai due to no lossless pg for some platforms Jan 10, 2025

nhe-NV approved these changes Jan 10, 2025

View reviewed changes

nhe-NV added the Request for 202411 branch label Jan 10, 2025

nhe-NV added Request for msft-202412 Branch and removed Request for 202411 branch labels Feb 19, 2025

JibinBao force-pushed the qos_no_pg_lossless branch from 3cd040a to abc7410 Compare February 19, 2025 08:14

mssonicbld added the Cherry Pick Conflict_msft-202412 label Feb 19, 2025

stephenxs approved these changes Feb 19, 2025

View reviewed changes

noaOrMlnx mentioned this pull request Feb 19, 2025

[Mellanox] Update DSCP mapping for SN5600, SN5610 SKUs sonic-net/sonic-buildimage#21762

Merged

11 tasks

mssonicbld mentioned this pull request Feb 20, 2025

[action] [PR:21762] [Mellanox] Update DSCP mapping for SN5600, SN5610 SKUs Azure/sonic-buildimage-msft#651

Merged

11 tasks

JibinBao added 4 commits March 5, 2025 10:43

update qos case due to no lossless buffer

6d3d0ec

1. When there is no lossless buffer, return a dump buffer lossless pg profile 2. Skip tests related to lossless buffer case dynamically

update the qos sai test for lossy queue according to the PR: sonic-ne…

0cae973

…t/sonic-buildimage#21762

add SN5640 to the margin setting list

80a72f8

JibinBao force-pushed the qos_no_pg_lossless branch from 1600d33 to 80a72f8 Compare March 5, 2025 08:48

kperumalbfn reviewed Mar 26, 2025

View reviewed changes

use cable_length 0m check instead of BUFFER_PG:4 as the comment

6a18cad

noaOrMlnx mentioned this pull request Mar 31, 2025

[Mellanox] Update SN56xx relevant SKUs buffers_defaults_objects.j2 sonic-net/sonic-buildimage#22067

Merged

11 tasks

liat-grozovik approved these changes Apr 1, 2025

View reviewed changes

liat-grozovik merged commit 485ec3a into sonic-net:master Apr 1, 2025
14 checks passed

JibinBao mentioned this pull request Apr 25, 2025

Update qos sai due to no lossless pg for some platforms Azure/sonic-mgmt.msft#225

Merged

12 tasks

r12f added Approved for msft-202412 Branch Included in msft-202412 Branch and removed Cherry Pick Conflict_msft-202412 labels Apr 28, 2025

mssonicbld added the Cherry Pick Conflict_msft-202412 label Apr 28, 2025

r12f removed the Cherry Pick Conflict_msft-202412 label Apr 28, 2025

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Update qos sai due to no lossless pg for some platforms #16440

Update qos sai due to no lossless pg for some platforms #16440

JibinBao commented Jan 10, 2025 •

edited by liat-grozovik

Loading

mssonicbld commented Jan 10, 2025

azure-pipelines bot commented Jan 10, 2025

JibinBao commented Jan 10, 2025

mssonicbld commented Jan 10, 2025

azure-pipelines bot commented Jan 10, 2025

mssonicbld commented Feb 19, 2025

azure-pipelines bot commented Feb 19, 2025

mssonicbld commented Feb 19, 2025

azure-pipelines bot commented Feb 19, 2025

mssonicbld commented Feb 21, 2025

azure-pipelines bot commented Feb 21, 2025

bingwang-ms commented Feb 26, 2025

kperumalbfn commented Feb 28, 2025

JibinBao commented Mar 5, 2025

mssonicbld commented Mar 5, 2025

azure-pipelines bot commented Mar 5, 2025

JibinBao commented Mar 17, 2025

bingwang-ms commented Mar 26, 2025

kperumalbfn Mar 26, 2025

JibinBao Mar 31, 2025

kperumalbfn Apr 1, 2025

mssonicbld commented Mar 31, 2025

azure-pipelines bot commented Mar 31, 2025

r12f commented Apr 7, 2025

r12f commented Apr 15, 2025

JibinBao commented Apr 16, 2025

r12f commented Apr 18, 2025 •

edited

Loading

JibinBao commented Apr 25, 2025

r12f commented Apr 28, 2025

Update qos sai due to no lossless pg for some platforms #16440

Update qos sai due to no lossless pg for some platforms #16440

Conversation

JibinBao commented Jan 10, 2025 • edited by liat-grozovik Loading

Description of PR

Type of change

Back port request

Approach

What is the motivation for this PR?

How did you do it?

How did you verify/test it?

Any platform specific information?

Supported testbed topology if it's a new test case?

Documentation

mssonicbld commented Jan 10, 2025

azure-pipelines bot commented Jan 10, 2025

JibinBao commented Jan 10, 2025

mssonicbld commented Jan 10, 2025

azure-pipelines bot commented Jan 10, 2025

mssonicbld commented Feb 19, 2025

azure-pipelines bot commented Feb 19, 2025

mssonicbld commented Feb 19, 2025

azure-pipelines bot commented Feb 19, 2025

mssonicbld commented Feb 21, 2025

azure-pipelines bot commented Feb 21, 2025

bingwang-ms commented Feb 26, 2025

kperumalbfn commented Feb 28, 2025

JibinBao commented Mar 5, 2025

mssonicbld commented Mar 5, 2025

azure-pipelines bot commented Mar 5, 2025

JibinBao commented Mar 17, 2025

bingwang-ms commented Mar 26, 2025

kperumalbfn Mar 26, 2025

Choose a reason for hiding this comment

JibinBao Mar 31, 2025

Choose a reason for hiding this comment

kperumalbfn Apr 1, 2025

Choose a reason for hiding this comment

mssonicbld commented Mar 31, 2025

azure-pipelines bot commented Mar 31, 2025

r12f commented Apr 7, 2025

r12f commented Apr 15, 2025

JibinBao commented Apr 16, 2025

r12f commented Apr 18, 2025 • edited Loading

JibinBao commented Apr 25, 2025

r12f commented Apr 28, 2025

JibinBao commented Jan 10, 2025 •

edited by liat-grozovik

Loading

r12f commented Apr 18, 2025 •

edited

Loading