[c-nsp] GigE sub-int won't come up

[Justin Shore]

Justin Shore wrote:
> Justin M. Streiner wrote:
>> Is there an active access port in VLAN 999 on the 4948, or somewhere 
>> downstream of it, assuming any neccessary trunking at the site is 
>> already in place?  I've seen VLANs not come up before, until there is 
>> actually a host in the VLAN.
> 
> There is an active switchport in 999 currently and 1001 before I 
> switched to 999.  I don't know that the 4948 could be the problem 
> though.  I've seen VLANs not come up until a port in the VLAN was active 
> as well but that's normally on a switch or a router with a switchport 
> (ISR w/ a Ethernet HWIC for example).  There isn't a mechanism to 
> advertise VLANs or their local status across a trunk that I'm aware of 
> (with VTP disabled at least).  VTP is transparent on the 4948 and not 
> configurable on the 7206.  I should be able to configure a dozen 1Q 
> sub-ints on a router's interface and have them be up/up regardless of 
> whether it's configured for use on the connected switch.  At least I'm 
> pretty sure I should be able to.

I had what was supposed to be a quick maintenance window tonight to bump 
the code rev on the 7200 and reboot.  I also did a minor rev update on 
the 4948.  I rebooted the 4948 first but after 10m it still hadn't come 
back up (I don't have OOB access to anything in that POP).  The 7200 
hadn't seen the interfaces come up.  I went ahead and did the 7200 while 
I was getting my things together to drive to that POP.  The 7200 never 
came back up either.  Joy.  Once I got onsite I found that both devices 
were in fact running.  The problem was that the GigE links on the 7200 
wouldn't come up.  Both the physical GigE interfaces and all their 
sub-ints were all up/down.  I started opening a TAC case at that point.

While the TAC operator waded through the tech options to try and figure 
out how to assign my case I thought about the problem some more, what 
Justin wrote earlier and my response about no VTP or VTP-like protocols 
in use jumped out at me.  I checked the config and sure enough I had CFM 
  and OAM (partially?) configured.  Don't they share link information 
including VLAN info?  I need to do more research on it and move the 
config to a lab environment.  As soon as I removed the CFM and OAM 
config from the 7200 the GigE links came up.  Not only that but all the 
sub-ints came up that I'd been fighting earlier.  My broken CFM or OAM 
or both config is what caused all these problems.

I was working with CFM and OAM after attending some MetroE training in 
SJC.  At the time my gear wasn't in production but now it is and I 
hadn't removed the config.  Does anyone have any good docs that clearly 
explain how to properly configure CFM, OAM and LMI?  I've found lots of 
docs that talk about it but none are terribly clear on exactly how to 
implement it and what's the BCP in certain environments.  Clearly what I 
was doing before wasn't right.  How to troubleshoot these protocols 
would also be very helpful.  Had I known that this is what was keeping 
the interface facing the 4948 down I could have fixed it from my 
recliner instead of taking a roadtrip to a remote POP.

Thanks for the info
  Justin