General Stack Configuration Guidelines
  
  
    Before deploying a new stack, consider the following guidelines:
      - Plan to use the stack as if it were a single multi-slot switch. You
          need to decide the number and type of stackable switches in the stack and how the stack
          ports will be connected to the network.
- Physically locate the intended master and backup nodes adjacent to
          each other, and plan to directly connect these nodes to each other so that ExtremeXOS
          application synchronization traffic is localized to a single stack link.
- Use stacking cables to interconnect the stack nodes into a ring
          topology (see SummitStack Topologies). Only
          include the nodes that are intended to be active in the stack. To see the recommended
          procedures for installing and interconnecting a stack, refer to the Summit Family Switches Hardware Installation Guide.
- You can physically connect the stack to your networks before the nodes
          are configured. However, the default configuration on a non-stacking mode switch assumes a
          default untagged VLAN that contains all switch ports. When first powered on, the switch
          acts as a Layer 2 switch, possibly resulting in network loops.
- Make sure all nodes support the SummitStack feature and are running
          the same ExtremeXOS software version (see the Feature License Requirements document.) To view the ExtremeXOS software version on a node, restart the node and run the
            command:
            show version {detail | process
              name | images {partition
              partition} {slot
              slotid} } .  If any of the nodes do not have the right version, install the correct version on that
            switch. We recommend that you use the same image partition on all nodes. Once stacking
            is enabled, image upgrade from the stack is possible only if the same image is selected
            on all nodes. 
- If you intend to deploy new units that might be part of a stack in the
          future, you might want to turn on stacking mode during initial deployment to avoid a
          future restart. The only disadvantages of stacking mode are the loss of QoS profile QP7
          and the reservation of some of the packet buffer space for stacking control traffic.
- You can configure the stack by logging into the master or any of the
          other nodes. For more information, see Logging into a Stack.
- If the master-capable stackable switches have different purchased
          license levels, you might need to configure license level restrictions on some nodes
          before those nodes can join the stack (see Managing Licenses on a Stack).
- If the stack supports any feature pack license (such as MPLS or Direct
          Attach), that feature pack license must be installed on all master capable nodes to
          support that feature and prevent traffic interruption if a failover event occurs.
- Most stacking specific configurations are effective only after a
          restart (see the following table). However, most non-stacking configuration commands take
          effect immediately and require no restart.
- A basic stack configuration can be achieved by using the procedure
          described in About Easy Setup.

Note   
If EAPS, Spanning Tree, or any Layer 2 redundancy
      protocol is not running on the network, make sure that your network connections do not form a
      network loop.