LACP problem between Extreme network stack and Juniper VCP.

[dilu]

Hi,

We are currently facing a problem with a port sharing configuration.
On one end we have a X670 stack (2 switches) and on the other end we have two
Juniper switches in a virtual chassis (VCP).

The config is pretty straight forward. Extreme:
enable sharing 1:37 grouping 1:37,2:37 algorithm address-based L3_L4 lacp

Juniper:
xe-0/1/0 {

description "008";

ether-options {

802.3ad ae2;

}

}

xe-2/1/0 {

description "009";

ether-options {

802.3ad ae2;

}

}

ae2 {

description "to Mica IT";

aggregated-ether-options {

lacp {

active;

periodic slow;

}

}

The problem with this configuration is that port 2:37 (connected to port xe-0/1/0) frequently drops out of the LAG without the interface going down:

12/12/2016 14:59:51.01 Slot-1: Remove port 2:37 from aggregator
12/12/2016 14:59:51.00 Slot-1: Port 2:37 is Down, remove from aggregator 1:37
12/12/2016 14:59:51.00 Slot-1: Port 2:37 link down - remote fault
12/12/2016 14:59:26.39 Slot-1: Add port 2:37 to aggregator
12/12/2016 14:58:59.20 Slot-1: Remove port 2:37 from aggregator
12/12/2016 14:57:26.14 Slot-1: Add port 2:37 to aggregator

The things i’ve tried to fix this problem:
- Created a new LAG (ports 1:36 and 2:36) > no difference.
- Swapped the optics > no difference the problem still persists on port 2:37.
- Swapped fibers > no difference the problem still persists on port 2:37 but problem to the other port (xe-2/1/0) on the Juniper side.

The problem always seem to be on the "second" port on the Extreme stack.
Does anybody has any idea what could go wrong in the scenario?

[Tripathy__Priya]

Dilu,

I could see already the ospf and lACP issues are dealt with the opened case with GTAC and assisted by my colleague Aleixo Gomes. Whereas the ospf issue found to be not with the extreme device though. Let us follow up the LACP issue with that opened case itself. But if you need here the broader audience to address the juniper site logs then as suggested by Stachal please do provide them here.

[Stachal]

What do the logs indicate on the Juniper? can you post those as well.

[Stachal]

This may not be the issue but something to check I only say this because I have seen this recently and it was the same very Odd behavior. LR Optics so your using single mode 1310nm... verify the fiber and patches are all Single mode(generally yellow not aqua).. You will get link running single mode over multimode fiber however you will have very unpredictable behavior. Just ran into an issue where the patch panel of fiber was mislabeled single mode when in fact it was multimode and everyone was scratching their heads as to why it would randomly drop.

Clean your fiber before you terminate it always.. single mode is more susceptible to dust. Also if you have OTDR test out the fiber to verify loss/bends/kinks. And one more thing to check is using single mode LR optics if the distance is short you may have to put a fiber optic attenuator in place to reduce reflection depending on the light level readings.

[Aleixo_Gomes]

Hi Dilu ,

case created 01269755.
please provide requested logs,