cancel
Showing results for 
Search instead for 
Did you mean: 

Uplink configuration between EXOS stack 5320 and VOSS cluster

Uplink configuration between EXOS stack 5320 and VOSS cluster

Tamaya
New Contributor

Hello,

Can someone please confirm whether the uplink configuration between a VOSS cluster and an EXOS 5320 stack is correct? (See attached diagram.)

For EXOS-to-EXOS, how should the uplinks be configured when they connect to a 5520 cluster used for VPEX?

During a production test of the VOSS cluster ↔ EXOS connection, we saw Centreon alerts indicating that some switches became unreachable, which forced us to disconnect the link. Our environment is large and mixed, with VOSS, EXOS, and ERS switches (SPB, MLT, SLT…).

Could this be related to the configuration spanning tree settings? If so, what are the best practices for spanning tree in this scenario?
Also, what troubleshooting steps would you recommend to isolate the issue?

Thanks for your help!

1 ACCEPTED SOLUTION

rose42snowden
New Contributor

Given your mixed environment (VOSS, EXOS, ERS) and the Centreon alerts, it's highly probable your connection issues are related to Spanning Tree Protocol (STP) settings, or a misconfiguration in the uplink aggregation. For VOSS clusters and EXOS 5320 stacks, Spanning Tree should generally be disabled on SMLT (VOSS) or shared group (EXOS) uplink ports to allow for sub-second failover, using LACP for aggregation where possible. On the VOSS side, enabling SLPP (Simple Loop Prevention Protocol) is recommended, while EXOS access ports should have SLPP-Guard enabled. When connecting EXOS to a 5520 cluster for VPEX, ensure consistent LAG/MLAG configurations. For troubleshooting, begin by verifying STP status on all uplink ports, checking LACP/MLAG configurations for consistency, examining interface counters for errors, and collecting detailed logs from both VOSS and EXOS devices.

Regards,
Rose

View solution in original post

3 REPLIES 3

Christoph
Contributor

Hi,

one thing I saw in your diagram: You had set LACP timer to short interval on VOSS.
The timer should be the same on both ends of the link. 
So on EXOS "configure sharing 1:53 lacp timeout short" has to be set, too.
Or leave defaults on both ends.

Regards
Christoph

Hi Christoph,

Thanks for your message. I have read that in a Fabric Attach environment, there is no need to configure uplinks on the EXOS side, as they are handled automatically by the fabric.
https://documentation.extremenetworks.com/exos_32.5/GUID-F140CBF4-6E25-4DA7-92EC-DD5B645E1218.shtml

Regards,

rose42snowden
New Contributor

Given your mixed environment (VOSS, EXOS, ERS) and the Centreon alerts, it's highly probable your connection issues are related to Spanning Tree Protocol (STP) settings, or a misconfiguration in the uplink aggregation. For VOSS clusters and EXOS 5320 stacks, Spanning Tree should generally be disabled on SMLT (VOSS) or shared group (EXOS) uplink ports to allow for sub-second failover, using LACP for aggregation where possible. On the VOSS side, enabling SLPP (Simple Loop Prevention Protocol) is recommended, while EXOS access ports should have SLPP-Guard enabled. When connecting EXOS to a 5520 cluster for VPEX, ensure consistent LAG/MLAG configurations. For troubleshooting, begin by verifying STP status on all uplink ports, checking LACP/MLAG configurations for consistency, examining interface counters for errors, and collecting detailed logs from both VOSS and EXOS devices.

Regards,
Rose

GTM-P2G8KFN