Loop in a Ring Topology with 3850 and 3650 switches.

xeum96700 · ‎06-05-2016

Hi,

I have a ring topology with 8 siwtches , a switch core 3850 a 3850 and 6 3650 .All switches are connected by a two- port etherchanel Tengiga .

The problem is that when you close the ring, a loop is created , and the entire network is coming down .

so now the ring is closed in the following manner , half of the switches are configured tengiga portchannel two interfaces , and the other half are only two trunks. Although the ring is closed , several interfaces in portchannel leave me to err -disable

But now , it is required that the entire ring portchannel lift with two interfaces to Tengiga .

It is that STP does not allow more than 7 jumps , so the more we have one of the loop occurs and err -disable , so if I use REP between the two switches 3850 , ring could work properly? or could put a portchannel in shutdown and if required to change the suth down on the interface with a ip sla, is this possible?

or anyone has any idea.

This is the configuration on the switches

interface Port-channel10
description MEQ-GUARD-IDF1-3650-1
switchport mode trunk
!
interface Port-channel20
description MEQ-IDF5-3650-1
switchport mode trunk
!
!
!
interface TenGigabitEthernet2/1/1
switchport mode trunk
channel-group 10 mode on
!
interface TenGigabitEthernet2/1/2
description
switchport mode trunk
channel-group 10 mode on
!
interface TenGigabitEthernet2/1/3
switchport mode trunk
channel-group 20 mode on
!
interface TenGigabitEthernet2/1/4
description
switchport mode trunk
channel-group 20 mode on
!

!
interface TenGigabitEthernet2/1/1
switchport mode trunk

!
interface TenGigabitEthernet2/1/2

switchport mode trunk

!
interface TenGigabitEthernet2/1/3
switchport mode trunk
channel-group 20 mode on
!
interface TenGigabitEthernet2/1/4
description
switchport mode trunk
channel-group 20 mode on

Philip D'Ath · ‎06-06-2016

You may be been better off using metro Ethernet switches to solve this problem.

First, make sure the 3850 is the root of all the VLANs, using something like:

spanning-tree vlan 1 priority 16384

Next I would make sure you are using rapid spanning tree. This is very important, as standard spanning tree may never converge or have trouble converging in this scenario. This needs to be done on every switch.

spanning-tree mode rapid-pvst

xeum96700 · ‎06-06-2016

Hi Philip,

The core switch is the Root of STP for all the switches and rapid-pvst is already enabled in all the switches.

Philip D'Ath · ‎06-06-2016

RTSP should not have the same issue of 7 switch hops, because it converges much more quickly than the default 20s timer.

When you get this "err-disabled", what are you seeing in the switch log? I think something else might be wrong.

xeum96700 · ‎06-08-2016

Hi Philip,

The log are %PM-4-ERR_DISABLE: channel-misconfig (STP) error detected on interface putting interface in err-disable state

But all the etherchannel are the same configuration.

https://supportforums.cisco.com/discussion/11221506/3750x-channel-misconfig-error-detected

Philip D'Ath · ‎06-08-2016

What software version are you running on the switches?

Are you sure the physical ports at the same as well, on both sides?

xeum96700 · ‎06-09-2016

the softwareversion are 03.07.03E in all the switches.

And Yes, I have the same configuration for the port chanels in all the switches.

Joseph W. Doherty · ‎06-09-2016

Have you tried increasing the diameter parameter?

xeum96700 · ‎06-09-2016

Hi Joseph

The maximum diameter is 7. I change the max adge from 20 to 30

I had channel mode on, on the portchannel so I will try to change this and put the portchannel with LACP and mode active.

Joseph W. Doherty · ‎06-06-2016

Disclaimer

The Author of this posting offers the information contained within this posting without consideration and with the reader's understanding that there's no implied or expressed suitability or fitness for any purpose. Information provided is for informational purposes only and should not be construed as rendering professional advice of any kind. Usage of this posting's information is solely at reader's own risk.

Liability Disclaimer

In no event shall Author be liable for any damages wha2tsoever (including, without limitation, damages for loss of use, data or profit) arising out of the use or inability to use the posting's information even if Author has been advised of the possibility of such damage.

Posting

It is that STP does not allow more than 7 jumps

That's not fully correct - it's just an IEEE recommendation to set the diameter to 7 as a default. If you have a ring of 8, have you tried setting the diameter parameter to 8?

if I use REP between the two switches 3850 , ring could work properly?

If REP is supported, it would be an alternative (perhaps better) to STP.

If you don't use REP, you might also consider using MSTP. (You might also consider using LACP.)

Philip D'Ath · ‎06-06-2016

I checked, and REP is just supported on the 3850's and 3650's using the latest bleeding edge code.

rafael_ormedo · ‎06-09-2016

Hi,

If your core is 2 switches 3850 and connected
logically as if they were one, and has switches 6
3650, the diameter of its network 7 thus supports
the factory default setting of Cisco. 8
switches are in default STP configuration?
Because of your question I think so. If you
You need to connect the last switch today, now? With
factory default settings, this ring topology
will work, I think you did not expect convergence time
network, because with the default setting will work but the convergence time will be higher,
because the switches select to the MAC value of physical ports.
To work this topology today, right now, you could do the following:
after working hours, turn off all switches, to the core,
and then connect as follows:

And also take the cables between the switches 3650/3 and 3650/4.

3850/1 - 3850/2 - connect simultaneously

After,

In 3850/1:

3650/1
3650/2
3650/3

In 3850/2:

3650/4
3650/5
3650/6

And after all are connected, put the cable between switches 3650/3 and 3650/4.

And expect the network convergence time, about 2 minutes if they are close, but can take up to 5 minutes

This is not recommandation Cisco, but will work!

Note that if any switch loses power on and off, when you turn on again, the network will be dead and will converge again after the convergence time, so you need to properly configure the switches.

This topology will work, but not properly, just put to the network becomes operational.

THEN, I suggest you set the STP ALL CISCO switches:

http://www.cisco.com/c/en/us/td/docs/switches/lan/catalyst2960/software/release/12-2_53_se/configuration/guide/2960scg/swstp.html
http://www.cisco.com/c/en/us/support/docs/lan-switching/spanning-tree-protocol/5234-5.html

And configure the port priority in PortChannel:

http://www.cisco.com/c/en/us/support/docs/lan-switching/spanning-tree-protocol/5234-5.html#s1
http://www.cisco.com/c/en/us/td/docs/switches/lan/catalyst2960/software/release/12-2_53_se/configuration/guide/2960scg/swstp.html

PS1. You can also connect the switches in other ways, but to wait for the time of convergence.

PS2: Please mark the question as answered if your query has been answered.

Thank you.