AFT global-filter was introduced openconfig/public#1441 and the initial testing READMEs were added in openconfig/featureprofiles#5217.
My understanding is that aft-summaries should reflect the total AFT entries in the HW FIB, not the filtered summary. Specifically, the Route Summary (per protocol route count in the FIB) should remain untouched, meaning it will still contain the counts related to all prefixes.
The YANG paths in question are:
/network-instances/network-instance[name]/afts/aft-summaries/ipv4-unicast/protocols/protocol[origin-protocol]/state/counters/aft-entries
/network-instances/network-instance[name]/afts/aft-summaries/ipv6-unicast/protocols/protocol[origin-protocol]/state/counters/aft-entries
To ensure this expected behavior is clear, we need to document this expectation directly in the public YANG model descriptions.
AFT global-filter was introduced openconfig/public#1441 and the initial testing READMEs were added in openconfig/featureprofiles#5217.
My understanding is that aft-summaries should reflect the total AFT entries in the HW FIB, not the filtered summary. Specifically, the Route Summary (per protocol route count in the FIB) should remain untouched, meaning it will still contain the counts related to all prefixes.
The YANG paths in question are:
/network-instances/network-instance[name]/afts/aft-summaries/ipv4-unicast/protocols/protocol[origin-protocol]/state/counters/aft-entries
/network-instances/network-instance[name]/afts/aft-summaries/ipv6-unicast/protocols/protocol[origin-protocol]/state/counters/aft-entries
To ensure this expected behavior is clear, we need to document this expectation directly in the public YANG model descriptions.