Home page logo
/

nanog logo nanog mailing list archives

Re: Flapping POS Interface on Frame-relay between a Juniper and Cisco
From: Righa Shake <righa.shake () gmail com>
Date: Thu, 8 Dec 2011 18:02:00 +0300

The interface finally stabalised.

This after performing segment by segment loop tests at SDH level.
We found errors which were sorted after which the service has been better
the random flaps have since disappeared.

Now dealing with random BGP cease notifications I receive from my upstream.

Thank you all for your assistance in helping solve this.

Regards,
Righa Shake




On Wed, Dec 7, 2011 at 12:17 AM, Scott Morris <swm () emanon com> wrote:

The mismatch problem of DCE/DTE should definitely indicate that your
PVCs aren't up.  But that shouldn't result in the high quantity of CRC
errors in the interface counters.  That should just result in your LMI
enquiry count increasing with LMI response sitting at zero.

I have to say I've never tried running frame-relay on a SONET
interface.  And if you're only using a single PVC there, I'd certainly
love to know WHY you're doing that!  :)

But setting one side to DCE so that it'll respond to LMI will certainly
make the frame-relay (L2) portion of things operate properly.  But if
you have something else occurring causing the CRCs along the way, then
that's not really going to help at all!

Did anything else coincide with you changing the encapsulation?

Scott



On 12/6/11 4:05 PM, Scott Weeks wrote:
Did Jeff's suggestion work?

: interface POS0/0/0
: frame-relay intf-type dce

If so, please let the list know, so when someone comes
across this thread while searching for the fix they can
figure it out without having to email the list.  If it
didn't help contact me off-line and I will be happy to
troubleshoot it with you.

scott





________________________________________
From: Righa Shake [righa.shake () gmail com]
Sent: Saturday, November 19, 2011 11:11 AM
To: afnog () afnog org
Subject: Flapping POS Interface on Frame-relay between a Juniper and
Cisco

Hi,

Am having a problem that is buffling.

I recently changed a POS link encapsulation from PPP to Frame-relay.
Since that time the POS interface keeps resetting from time to time.

On my BGP session am receiving cease notifications from my upstream
provider.

The setup is such that we have a cisco on one end and a Juniper on the
other.

interface POS0/0/0
 mtu 4474
 no ip address
 no ip unreachables
 encapsulation frame-relay
 logging event link-status
 crc 32
 pos scramble-atm
 frame-relay lmi-type ansi
end

ROUTERshow run int pos0/0/0.101
Building configuration...


!
interface POS0/0/0.101 point-to-point
ip address X.X.X.X 255.255.255.252
frame-relay interface-dlci 101
end

ROUTER#show int pos0/0/0
POS0/0/0 is up, line protocol is up
  Hardware is SPA-2XOC12-POS
  MTU 4474 bytes, BW 622000 Kbit/sec, DLY 100 usec,
     reliability 255/255, txload 6/255, rxload 38/255
  Encapsulation FRAME-RELAY, crc 32, loopback not set
  Keepalive set (10 sec)
  Scramble enabled
  LMI enq sent  81981, LMI stat recvd 77480, LMI upd recvd 0, DTE LMI up
  LMI enq recvd 0, LMI stat sent  0, LMI upd sent  0
  LMI DLCI 0  LMI type is ANSI Annex D  frame relay DTE  segmentation
inactive
  FR SVC disabled, LAPF state down
  Broadcast queue 0/256, broadcasts sent/dropped 26/0, interface
broadcasts
0
  Last input 00:00:00, output 00:00:00, output hang never
  Last clearing of "show interface" counters 1w2d
  Input queue: 0/375/0/0 (size/max/drops/flushes); Total output drops: 0
  Queueing strategy: fifo
  Output queue: 0/40 (size/max)
  5 minute input rate 94336000 bits/sec, 13151 packets/sec
  5 minute output rate 16470000 bits/sec, 7049 packets/sec
     12211574207 packets input, 10967607038364 bytes, 0 no buffer
     Received 0 broadcasts (0 IP multicasts)
     6970870 runts, 2179 giants, 0 throttles
              0 parity
     892493293 input errors, 882184781 CRC, 0 frame, 0 overrun, 0
ignored,
3335463 abort
     6379191154 packets output, 1614018181446 bytes, 0 underruns
     0 output errors, 0 applique, 4 interface resets
     0 unknown protocol drops
     0 output buffer failures, 0 output buffers swapped out
     0 carrier transitions


Any assistance on this will be greatly appreciated.






--- jsaxe () briworks com wrote:

From: Jeff Saxe <jsaxe () briworks com>

I believe this is the explanation for your flapping: a PPP link is
intended to go between two routers over, for instance, a private leased
line, so both of the devices are peers, neither one particularly special.
Frame-relay, by contrast, was originally designed so that your router was
an "end user" device and its directly-connected partner device was not your
other router, which you control, but the frame carrier's frame-relay
switch. Your router was a DTE device, and their switch, which was in a more
"important" position in control of the frame-relay NBMA cloud, was the DCE
device. Your router then slaved to the frame switch via LMI signaling, so
that the upstream switch instructed you which DLCIs existed and were up at
the moment.

So if you connect up two routers with frame-relay encap and each thinks
it is the DTE, and neither one is taking the role of the frame switch, then
when you bring them up, they will initially optimistically think their
DLCIs are up and working, and the routing protocol and traffic will come
up... but both of them will be waiting for the frame switch to send them
LMI indicating that their idea of the DLCI up/down status is correct. When
a couple minutes go by and they don't hear the responses to their LMI
enquiries, they will bring all the DLCI's down. I thought they would then
stay down forever, i.e., not flap, but maybe you are shutting / no shutting
the POS circuit to try again. Anyway, I believe the very simple fix is

interface POS0/0/0
frame-relay intf-type dce

So this will turn your Cisco side of the circuit into "DCE" mode, and if
the Juniper side stays in "DTE" mode (the default, so probably not listed
in the config), then the LMI should start behaving between the two. And
yes, as Jay Hennigan suggested, you might need to use "encap frame-relay
ietf" to be compatible with non-Cisco gear, or you might need to adjust the
frame-relay lmi-type -- one type sends the LMI on DLCI number 0, one of
them on DLCI 1023, whatever. I think you'll need to adjust the two ends
until you see LMI enquiries both sent and received; right now the "show
interface" from the Cisco side shows it has not received any LMI enq yet.


Good luck, and I hope it's that simple.   :-)








  By Date           By Thread  

Current thread:
[ Nmap | Sec Tools | Mailing Lists | Site News | About/Contact | Advertising | Privacy ]