Port is down, reason:notconnect(connected), used by:Discovery

[michelvankessel]

Hello,

We have a ACI fabric running 1.1.1o. We have a FEX connected and applied a link policy for the 102/1/41-48 to be fixed 100Mb. 

Most ports are in use as you can see in the output below. 

LEAF2# show interface ethernet 102/1/41-58 brief 
--------------------------------------------------------------------------------
Ethernet      VLAN    Type Mode    Status Reason                   Speed     Port
Interface                                                                    Ch # 
--------------------------------------------------------------------------------
Eth102/1/41   65      eth  trunk   up     none                     100(D)    -- 
Eth102/1/42   84      eth  trunk   up     none                     100(D)    -- 
Eth102/1/43   0       eth  trunk   down   link-failure             100(D)    -- 
Eth102/1/44   82      eth  trunk   up     none                     100(D)    -- 
Eth102/1/45   82      eth  trunk   up     none                     100(D)    -- 
Eth102/1/46   82      eth  trunk   up     none                     100(D)    -- 
Eth102/1/47   0       eth  trunk   down   link-failure             100(D)    -- 
Eth102/1/48   76      eth  trunk   up     none                     100(D)    -- 

Ports Eth102/1/43 and Port Eth102/1/47 are both not part of an EPG. Port Eth102/1/47 has been used in an EPG before, port Eth102/1/43 is still unused.

If we look in the fault section in the APIC there is a warning 

"topology/pod-1/node-102/sys/phys-[eth102/1/47]/phys
Port is down, reason:notconnect(connected), used by:Discovery"

We don't get the same warning for interface  Eth102/1/43, but the port is in the same state. Is this normal behavior?

It looks like the warning only appears after you use a interface once and then you have to manually disable it to not affect the Fabric Health anymore. Any suggestions? 

thanks!

Michel van Kessel

[Zach Seils · Accepted solution]

Hi Michel,

Can you please check the Last Link St Change value for interfaces Eth102/1/43 and Eth102/1/47?  The presence of a valid timestamp is what controls whether a fault is raised for non-infrastructure ports.  If an interface has never come up a fault is not raised.  However, if the interface has come up, and the subsequently goes down, a fault is raised.  You can remove the fault by shutting down the interface.

Regards,

Zach

[michelvankessel]

Hello Zach,

I was already imagining that this would be the case. Just needed a confirmation

The default timestamp is 1970-01-01T02:00:00.000+02:00

After changing the status is will show a more recent date 2015-08-19T07:59:22.583+02:00

Many thanks

Michel