Skip to content

[Mellanox][Dual-Tor][DSCP remapping]Fix a test issue for the dualtor tunnel monitor for Nvidia platforms#9625

Merged
bingwang-ms merged 1 commit intosonic-net:masterfrom
congh-nvidia:tunnel_monintor
Jan 16, 2024
Merged

[Mellanox][Dual-Tor][DSCP remapping]Fix a test issue for the dualtor tunnel monitor for Nvidia platforms#9625
bingwang-ms merged 1 commit intosonic-net:masterfrom
congh-nvidia:tunnel_monintor

Conversation

@congh-nvidia
Copy link
Copy Markdown
Contributor

@congh-nvidia congh-nvidia commented Aug 24, 2023

Description of PR

Summary:
Fixes # (#9624)
On Nvidia platforms, when the inner/outer DSCPs are 2/2 or 6/6, the tunnel packet is not mapped based on the mapping configuration to queue 1 (they will be mapped to queue 2 and 6).
And we have confirmed with MSFT that such packets are not expected in production.
So, skip the queue check for them.

Type of change

  • Bug fix
  • Testbed and Framework(new/improvement)
  • Test case(new/improvement)

Back port request

  • 201911
  • 202012
  • 202205
  • 202305

Approach

What is the motivation for this PR?

Fix test issue (#9624)

How did you do it?

Skip the queue check in tunnel monitor on Nvidia platforms when dscp is 2/2 or 6/6

How did you verify/test it?

Run test on 4600C dualtor setup, passed.

Any platform specific information?

Only for Nvidia platforms.

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

Documentation

When the inner/outer dscps are 2/2 or 6/6, the packet is not mapped to queue according to the mapping configuration. And such packets are not expected in production, skip the queue check for them.

Change-Id: I8f02d086d2f9eb010684e1ac41cf1abfe6ca2788
@liat-grozovik
Copy link
Copy Markdown
Collaborator

@bingwang-ms kindly reminder. the test fix should be part of 202305.

@liat-grozovik liat-grozovik changed the title [Nvidia][Dual-Tor][DSCP remapping]Fix a test issue for the dualtor tunnel monitor for Nvidia platforms [Mellanox][Dual-Tor][DSCP remapping]Fix a test issue for the dualtor tunnel monitor for Nvidia platforms Oct 9, 2023
@bingwang-ms
Copy link
Copy Markdown
Collaborator

If the inner DSCP is 2 or 6, the outer DSCP should be 0. Is that right? Why the tunnel_monitor doesn't work?

DSCP_TO_TC_MAP:

"DSCP_TO_TC_MAP": {
        "AZURE": {
            ......
            "1": "1",
            "0": "1",
            "3": "3",
            "2": "1",
            "5": "1",
            "4": "4",
            "7": "1",
            "6": "1",
            "9": "1",
            "8": "0",
            "51": "1",
}

TC_TO_DSCP_MAP:

 "TC_TO_DSCP_MAP": {
        "AZURE_TUNNEL": {
            "1": "0",
            "0": "8",
            "3": "2",
            "2": "2",
            "5": "46",
            "4": "6",
            "7": "48",
            "6": "6",
            "8": "33"
        }
    }

@congh-nvidia
Copy link
Copy Markdown
Contributor Author

Hi @bingwang-ms ,
On Nvidia platforms, we have different implementation in dscp remapping. The inner dscp 2 and 6 are considered as invalid use cases and not handled according to the mapping configurations.

@liat-grozovik
Copy link
Copy Markdown
Collaborator

@bingwang-ms kindly reminder, this fix is needed also for 202305

@congh-nvidia
Copy link
Copy Markdown
Contributor Author

Hi @bingwang-ms , could you approve the PR or let me know if you still have any concerns?
Thanks.

@bingwang-ms
Copy link
Copy Markdown
Collaborator

@congh-nvidia Sorry I missed the review. So what's the outer DSCP value if the inner DSCP is 2 or 6?

@congh-nvidia
Copy link
Copy Markdown
Contributor Author

Hi @bingwang-ms , on Nvidia platforms the inner dscp 2/6 will be encapsulated with outer dscp 2/6 on the standby tor.

@bingwang-ms bingwang-ms merged commit 1e4cb0c into sonic-net:master Jan 16, 2024
mssonicbld pushed a commit to mssonicbld/sonic-mgmt that referenced this pull request Jan 17, 2024
…onic-net#9625)

When the inner/outer dscps are 2/2 or 6/6, the packet is not mapped to queue according to the mapping configuration. And such packets are not expected in production, skip the queue check for them.

Change-Id: I8f02d086d2f9eb010684e1ac41cf1abfe6ca2788
@mssonicbld
Copy link
Copy Markdown
Collaborator

Cherry-pick PR to 202311: #11299

mssonicbld pushed a commit to mssonicbld/sonic-mgmt that referenced this pull request Jan 17, 2024
…onic-net#9625)

When the inner/outer dscps are 2/2 or 6/6, the packet is not mapped to queue according to the mapping configuration. And such packets are not expected in production, skip the queue check for them.

Change-Id: I8f02d086d2f9eb010684e1ac41cf1abfe6ca2788
@mssonicbld
Copy link
Copy Markdown
Collaborator

Cherry-pick PR to 202305: #11300

mssonicbld pushed a commit that referenced this pull request Jan 17, 2024
…9625)

When the inner/outer dscps are 2/2 or 6/6, the packet is not mapped to queue according to the mapping configuration. And such packets are not expected in production, skip the queue check for them.

Change-Id: I8f02d086d2f9eb010684e1ac41cf1abfe6ca2788
mssonicbld pushed a commit that referenced this pull request Jan 22, 2024
…9625)

When the inner/outer dscps are 2/2 or 6/6, the packet is not mapped to queue according to the mapping configuration. And such packets are not expected in production, skip the queue check for them.

Change-Id: I8f02d086d2f9eb010684e1ac41cf1abfe6ca2788
@congh-nvidia congh-nvidia deleted the tunnel_monintor branch October 17, 2024 02:11
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Projects

None yet

Development

Successfully merging this pull request may close these issues.

7 participants