This question is for GigaBlox SFP, but the answer I provide below applies pretty much to all our hardware, and probably to any ethernet hardware in general (though I can’t speak for hardware we didn’t design).
In general, when water gets onto a device, you can’t make any useful predictions about failure modes. Water gets everywhere, and if it’s present across the pins of the ethernet ports, it’s probably also present elsewhere on the board.
That being said, let’s assume that the water only gets onto the ethernet pins of one port.
What happens now depends on the kind of water. Pure water is not a good conductor and thus it probably won’t have any effect on the ethernet port. Of course, it’s unlikely to be pure water. Saltier water like sea water conducts electricity, fresh water is less saline but still conducts. Let’s assume this is sea water.
Sea water will provide a conduction path between the pins of the ethernet port. In the worst case, this will short circuit the pins. If it short circuits any of the pins A+, A-, B+ or B-, then the whole port will go down. If it only shorts any of the pins C+, C-, D+ or D-, then the port will downgrade from 1000BASE-T to 100BASE-T. This is because only pins A+, A-, B+ and B- are needed to carry the 100BASE-T data, and ethernet’s auto-negotiation capability will detect if only the A and B are present, and downgrade connection speed to 100BASE-T.
However, it is possible that sea water provides some conduction between the pins but does not fully short circuit them. In this case, the port might stay active, but signal integrity is reduced. You might not even notice a connection issue in this case.
In any case, if the water is only touching the ethernet pins of one port, none of the the ports will be affected, and the device will function as normal (apart from the affected port). Just remember what I said regarding water though; if it’s somewhere on the board, it’s probably other places too. Sea water across the power pins would probably cause the switch to stop functioning.