Ethernet (BaseT) down: poll driven or event driven?

fausta.t · ‎11-24-2009

Hi All,

when detaching a cable from a BaseT port there is a certain delay before the "interface down" message shows up in the log . Does this explain with the fact that according to IEEE standards the "interface down" status for BaseT interfaces is poll driven instead of event driven (like, for example, for fiber GigEth ports) ?

If so, what is the polling interval on Cisco switches?

Thanks in advance.

Best Regards.

Fausta.

Giuseppe Larosa · ‎11-25-2009

Hello Fausta,

ethernet has no carrier or frame structure to detect, so the ethernet port has to detect that its own keepalive frames are not coming back to detect that the cable has been unplugged.

if on an RJ-45 port of a router for example you issue no keepalive the port is fooled to up/up even without any cable plugged in!

And it can be pinged because LAN interface doen't send the icmp packet on wire as serial or ATM intefaces do.

I used this trick some years in some test labs.

so the delay should be this time of missing some keepalives in a row.

Hope to help

Giuseppe

fausta.t · ‎11-26-2009

Hello Giuseppe,

thank you for your answer, my question though was a little different, I'll try to make myself clearer . Let's say the event is "I missed some keepalives in a row" and this happens at T0: Is the "interface down" message displayed at T0 (event driven) or do we have to wait for a polling process to check the status of all the interfaces and declare the interface down (poll driven)? In this case the "interface down" message would be displayed at T0+the time needed for the polling process to verify the interface status.

Thank you.

Cheers

Fausta.

Giuseppe Larosa · ‎11-26-2009

Hello Fausta,

according to my experience I think it should happen at T0 so event driven for using your terminology.

Hope to help

Giuseppe