I have a Frame connection between two sites, A and B. If the router
crashes at B, wouldn't A still see the DLCI for
B? Is there any scenario where this wouldn't be the case? If
B gets blown off the map, shouldnt A's frame interface be
in at least an up/down scenario?
The only reason I ask is because AT&T has been working on a
17hr outage like this and they are focusing on the CO at site B. When I
asked them how clearing up a CO issue would make the DLCI
reaappear at site A, they didn't have a straight answer.
In my particular case, the remote site B actually has a PVC to
A and C and both of those are in a down/down....these are
subinterfaces on DS3's with tons of other sites working properly.
If the DLCI disappeared, doesn't that suggest that someone
deleted the PVC?
Any insights would be helpful....
Thanks,
-BM
PS It would be reasonable to interpret this *partially* as a rant.
I'd really like to confirm the PVC vs CO scope of the problem
though.
Thus spake <Brennan_Murphy@NAI.com>
I have a Frame connection between two sites, A and B. If the router
crashes at B, wouldn't A still see the DLCI for
B? Is there any scenario where this wouldn't be the case? If
B gets blown off the map, shouldnt A's frame interface be
in at least an up/down scenario?
In theory, when the remote LMI goes down, or any part of the PVC internal to the
carrier goes down, the local LMI should go INACTIVE. In practice, this is
unreliable at best; for some carriers, you will always get ACTIVE status no
matter what.
ATM has a similar feature/problem with ILMI. The only solution is to ignore the
carrier and do it yourself end-to-end (gosh, where have we heard that before?).
End-to-End Keepalives for FR:
http://www.cisco.com/univercd/cc/td/doc/product/software/ios121/121cgcr/wan_r/wr
dfrely.htm#xtocid18
OAM Loopback for ATM:
http://www.cisco.com/univercd/cc/td/doc/product/software/ios120/12cgcr/wan_r/wro
ampvc.htm#xtocid0
If the DLCI disappeared, doesn't that suggest that someone
deleted the PVC?
Distinguish between INACTIVE and DELETED; those represent very different things
in LMI.
S
> I have a Frame connection between two sites, A and B. If the router
> crashes at B, wouldn't A still see the DLCI for
> B? Is there any scenario where this wouldn't be the case? If
> B gets blown off the map, shouldnt A's frame interface be
> in at least an up/down scenario?
In theory, when the remote LMI goes down, or any part of the PVC internal to the
carrier goes down, the local LMI should go INACTIVE. In practice, this is
unreliable at best; for some carriers, you will always get ACTIVE status no
matter what.
For a real frame relay (not the frame relay on customer end that terminates
into the ATM circuit on your end) if you want the interface to go up/down or
down/down when there is a problem on a line, you must configure frame relay
using sub-interfaces. Works like a charm.
Alex
Thus spake <alex@yuriev.com>
> In theory, when the remote LMI goes down, or any part of the PVC internal to
the
> carrier goes down, the local LMI should go INACTIVE. In practice, this is
> unreliable at best; for some carriers, you will always get ACTIVE status no
> matter what.
For a real frame relay (not the frame relay on customer end that terminates
into the ATM circuit on your end) if you want the interface to go up/down or
down/down when there is a problem on a line, you must configure frame relay
using sub-interfaces. Works like a charm.
With some carriers, that works. With others, it doesn't. If you want to count
on it working 24x7 with any carrier, you need end-to-end checks.
S