At a water treatment plant, one physical member of an EtherChannel fails while the remaining members stay up. What should operators expect for traffic that still hashes to healthy members?

When one physical member of an EtherChannel fails at a water treatment plant, the Port-Channel remains up as long as at least one member is still active. Traffic whose hash selects healthy members continues to forward on those members. Operators should expect reduced aggregate capacity and a redistribution of some flows that previously hashed to the failed link, not a total outage of the logical interface. That partial-capacity behavior is a primary operational benefit of bundling versus depending on a single physical uplink.

The idea that the entire Port-Channel must go down until every member is restored is incorrect for standard EtherChannel operation with remaining active members. Spanning tree immediately blocking all remaining members for thirty minutes is not the default reaction to a single member failure; STP sees the Port-Channel as one logical link whose state tracks the bundle, not a mandate to quarantine survivors. Automatic conversion of all VLANs on the Port-Channel to access mode has no basis in EtherChannel failure mechanics.

Exam caveat: if the bundle drops below a configured minimum-links threshold, the Port-Channel can be forced down deliberately—know whether minimum links is set in the plant design. Also, flows that hashed to the failed member move only after the hash membership changes; brief disruption for those flows is expected even though the channel stays up.

Operational check: fail one member in a maintenance window, confirm Port-Channel stays up, verify counters move to remaining members, check whether minimum-links would have dropped the bundle, and restore the member while watching LACP/ON status return cleanly.