[nsp] MLPPP problems in 12.0(25)S1?

Raymond, Steven steven_raymond at eli.net
Tue Aug 19 12:30:40 EDT 2003

So since upgrading to 12.0(25)S1 from 12.0(21)S5, have suffered several
interruptions on a 7-T1 MLPPP bundle with a PA-MC-T3 VIP2-50/RSP4.
While this particular bundle has been problematic in the past, (21)S5 seemed
impressively stable.
Now when this latest failure occurs, there are no syslog events, no up/down
transitions, not a hiccup.  The directly-connected MLPPP neighbor becomes
unpingable (passing of packet traffic ceases), all the while layer 1 & 2 are
happy as can be.  "show int multilinkX" indicates that for every packet
received, input queue drop counter increases by one.  Tried the following
with no success:

1)  Bounced the serial and multilink interfaces
2)  Delete, deconfigure & reconfigure the serial and multilink interfaces
3)  Reloaded the entire vip slot

The following workarounds temporarily fix it, for as short as say 24 hours
and as long as a week:

1)  Reload the whole router
2)  Recreate a new multilink interface and simply move the T1s to

Another strange indication is that when the input queue drops are happening,
the "show int multilinkX" has 0* throttles in the output.  Have never seen
an asterisk appear in the output of a show interface command.

Opened a TAC case and once a workaround was found, they became uninterested
in diagnosing the root cause.  Has occurred three times so far and I am
tired of it already.

For that matter- how well does MLPPP work on a 6CT3-SMB in a GSR?  Any
positive or negative experiences out there?  Am thinking about moving this
bundle onto a GSR.

Thank you

More information about the cisco-nsp mailing list