[202405][GCU] ERR log - "Unsupported port <intf> speed 100000" seen during speed-lane change

[amitpawar12]

Is it platform specific

generic

Importance or Severity

High

Description of the bug

The GCU script is trying to configure 100Gbps port on multi-asic LC of 400G HWSKU.

The GCU script is removing all QoS, Buffer related configuration from the port, admin-state down, changes the lanes, speed and fec on the port before bringing it up.

The GCU config apply-patch successfully goes through, and the speed and lane of the port is also changed. Hence severity is set to HIGH for this issue.

We have filed an issue where the speed and port changes causes incorrect re-addition of the system ports. Even with the fix for the same, we are still seeing this ERR log.

@saksarav-nokia @rawal01 @okaravasi for viz.

Before GCU config:

admin@ixre-egl-board74:/etc/sonic$ show interface status
  Interface                            Lanes    Speed    MTU    FEC         Alias    Vlan    Oper    Admin                                             Type    Asym PFC
-----------  -------------------------------  -------  -----  -----  ------------  ------  ------  -------  -----------------------------------------------  ----------
 Ethernet40  112,113,114,115,116,117,118,119     400G   9100    N/A   Ethernet6/1  routed    down     down                                  QSFP28 or later         off

Following is the GCU script:

[
{ "op": "remove",
  "path": "/asic0/BUFFER_QUEUE/ixre-egl-board74|asic0|Ethernet40|0-2"
},
{ "op": "remove",
  "path": "/asic0/BUFFER_QUEUE/ixre-egl-board74|asic0|Ethernet40|3-4"
},
{ "op": "remove",
  "path": "/asic0/BUFFER_QUEUE/ixre-egl-board74|asic0|Ethernet40|5-6"
},
{ "op": "remove",
  "path": "/asic0/QUEUE/ixre-egl-board74|asic0|Ethernet40|0"
},
{ "op": "remove",
  "path": "/asic0/QUEUE/ixre-egl-board74|asic0|Ethernet40|1"
},
{ "op": "remove",
  "path": "/asic0/QUEUE/ixre-egl-board74|asic0|Ethernet40|2"
},
{ "op": "remove",
  "path": "/asic0/QUEUE/ixre-egl-board74|asic0|Ethernet40|3"
},
{ "op": "remove",
  "path": "/asic0/QUEUE/ixre-egl-board74|asic0|Ethernet40|4"
},
{ "op": "remove",
  "path": "/asic0/QUEUE/ixre-egl-board74|asic0|Ethernet40|5"
},
{ "op": "remove",
  "path": "/asic0/QUEUE/ixre-egl-board74|asic0|Ethernet40|6"
}, 
{ "op": "remove",
  "path": "/asic0/BUFFER_PG/Ethernet40|0"
},
{ "op": "remove",
  "path": "/asic0/PORT_QOS_MAP/Ethernet40"
},
{
    "op": "add",
    "path": "/asic0/PORT/Ethernet40/admin_status",
    "value": "down"
},
{
    "op": "add",
    "path": "/asic0/PORT/Ethernet40/lanes",
    "value": "112,113,114,115"
},
{
    "op": "add",
    "path": "/asic0/PORT/Ethernet40/speed",
    "value": "100000"
},
{
    "op": "add",
    "path": "/asic0/PORT/Ethernet40/fec",
    "value": "none"
},
{
    "op": "add",
    "path": "/asic0/PORT/Ethernet40/admin_status",
    "value": "up"
}
]

This ends with following ERR log:

2025 Nov  6 17:21:45.556638 ixre-egl-board74 ERR swss0#orchagent: :- doPortTask: Unsupported port Ethernet40 speed 100000
2025 Nov  6 17:21:46.964443 ixre-egl-board74 ERR swss0#orchagent: :- doPortTask: Unsupported port Ethernet40 speed 100000

However, the lane and speed change are successful:

admin@ixre-egl-board74:~$ show interface status
  Interface                            Lanes    Speed    MTU    FEC         Alias    Vlan    Oper    Admin                                             Type    Asym PFC
-----------  -------------------------------  -------  -----  -----  ------------  ------  ------  -------  -----------------------------------------------  ----------
 Ethernet40                  112,113,114,115     100G   9100   none   Ethernet6/1  routed      up       up                                  QSFP28 or later         off

Steps to Reproduce

1. Load minigraph with no configuration set on 400G port on 400G HWSKU.
2. Add gcu patch to change the lanes and speed of the interface.
3. Monitor for ERR level logs.

Actual Behavior and Expected Behavior

Expected behavior:
No ERR log should be generated during this process.

Actual behavior:
Although GCU patch goes through successfully, the ERR level log is generated.

Relevant log output

Output of show version, show techsupport

Attach files (if any)

No response

[saksarav-nokia]

I will investigate this further with @amitpawar12

[prgeor]

@bhouse-nexthop offered help to make GCU smart enough to understand the patch to decide the right sequencing and do all the removal of port attributes followed by port removal and finally the addition of the port attribute

[arlakshm]

This issue is track the fix in master. For 202405, this log message will be ignored as there is not functional impact

[amitpawar12]

This log will be ignored in the sonic-mgmt GCU port speed-lane change testcases.

Keeping the issue open to check on master branch.

Thanks.

[bhouse-nexthop]

Right now RemoveCreateOnlyDependencyMoveValidator generates really bad plans, and since the lanes are changing it would flow through that generator. It would have been good to see the log messages generated by GCU on its actual calculated plan to confirm if it wasn't doing things right.

I'm opening a PR in the next couple of days that completely reworks RemoveCreateOnlyDependencyMoveValidator to fix the plan generation. I'll link it here once I create it.

[bhouse-nexthop]

This is likely fixed by: sonic-net/sonic-utilities#4335 if it does not work, please apply your patch with --path-trace /tmp/trace.json and provide the generated file (NOTE: path trace was added in sonic-net/sonic-utilities#4317).

Also, I should mention your patch does a lot of unnecessary work. There is no need to set admin_status down and back up as GCU applies all the patches to determine the final state then creates its own path.