Unset SAI_PORT_ATTR_HW_LANE_LIST as key

[byu343]

The reason is that we are using one syncd process to support multiple phy chips and for the ports from different phys, they may
have the same value on SAI_PORT_ATTR_HW_LANE_LIST.

This change is depended by sonic-net/sonic-buildimage#8146. Specifically, multiple switches are defined under guid 1 as below, however, ports from different switches may have the value on SAI_PORT_ATTR_HW_LANE_LIST.
{
"guid" : 1,
"name" : "phys", "dbAsic" : "GB_ASIC_DB",
"dbCounters" : "GB_COUNTERS_DB",
"dbFlex": "GB_FLEX_COUNTER_DB",
"dbState" : "STATE_DB",
"zmq_enable" : false,
"zmq_endpoint": "tcp://127.0.0.1:5565",
"zmq_ntf_endpoint": "tcp://127.0.0.1:5566",
"switches": [
{
"index" : 0,
"hwinfo" : "mdio0_0_0/0"
},
{
"index" : 1,
"hwinfo" : "mdio1_0_0/0"
}
}

Signed-off-by: Boyang Yu [email protected]

[lguohan]

i do not understand this, they are different switches, the key should only matter on one switch. where do you see the issue?

[kcudnik]

hw lane list identifies uniquely ports on a single switch, it must be a KEY, please revert

[kcudnik]

is the error you getting from metadata in OA or syncd? can you paste error? is the scenario you are facing is that you have 2 SAI switches, one of the ports on 1st switch and one of the ports on 2nd switch have the same lanes ?

[kcudnik]

this fix will resolve that: sonic-net/sonic-sairedis#865

[byu343]

Sure. I'll the try your fix and hopefully close this PR thereafter. Thanks.

[byu343]

i do not understand this, they are different switches, the key should only matter on one switch. where do you see the issue?

Yes, the ports having the same SAI_PORT_ATTR_HW_LANE_LIST are from different switches.

The error is
Jul 20 16:03:41.998421 smm211 DEBUG swss#orchagent: :- constructKey: constructed key: SAI_PORT_ATTR_HW_LANE_LIST:2,3;
Jul 20 16:03:41.998421 smm211 ERR swss#orchagent: :- meta_generic_validation_create: attribute key SAI_PORT_ATTR_HW_LANE_LIST:2,3; already exists, can't create_

raised from https://github.com/Azure/sonic-sairedis/blob/216e5497bfb23e697c269fdac0d6965902ebe644/meta/Meta.cpp#L5060.

It is happening under the setup of using one syncd process and one context to serve 8 phys. I'll investigate further why it happens.

[kcudnik]

my fix on sairedis meta addresses this issue, you can close this issue here, and move it to sairedis issues, since related to sairedis not SAI

[byu343]

The issue is fixed by sonic-net/sonic-sairedis#865.