Troubleshooting a Stack

Use this section to diagnose and troubleshoot common configuration errors for stacking.

The most common errors are:
  • The stack did not come up as expected: Use the show stacking, show stacking configuration, show stacking-support, and show stacking stack-ports commands to diagnose the problem. There could be incorrect stacking cabling, a configuration error, or powered down nodes. Also, check the log using the show log command.

  • The switch with the highest priority was not elected manager: nodes might have been powered up at different times. Reboot all nodes in the stack simultaneously.

  • A node appears in the stack as expected but does not appear to be operating as configured: Use the show slot {slot {detail} | detail } command to see if there is a license mismatch or an incorrect ExtremeXOS software version. For more information, see Managing Licenses on a Stack.

  • A correctly cabled and powered-on node does not appear in the stack: The node might be running an ExtremeXOS version that is earlier than ExtremeXOS 12.0. Upgrade the ExtremeXOS version using the procedure you would use if the node was not part of the stack.

  • The following message appears: Warning: The Backup stack node is not as powerful or as capable as the Master stack node. This configuration is not recommended for successful use of the failover feature.

    This message appears once, about five minutes after master node takes control of the stack. To better support the failover feature, follow the guidelines in Configuring the Master, Backup, and Standby Roles.

  • The following message appears: Notice: There are Standby stack nodes which are more powerful and more capable than the Master and/or Backup stack nodes. This configuration is not recommended for optimal stack performance. We recommend that you reconfigure the stacking master-capability and/or priority parameters to allow the higher performing and more capable nodes to become Master and/or Backup stack nodes.

    This message appears once, about five minutes after master node takes control of the stack. To optimize stack performance, follow the guidelines in Configuring the Master, Backup, and Standby Roles.

  • All saved files (backup configurations, script, etc.) are lost on node that becomes slave when implementing stacking.
    Note

    Note

    Disabling or deleting the stacking configuration does not restore the files.
  • Either or both ends of a stacking link show that the state is No Neighbor. This can happen when the port at either end is configured incorrectly or if there is a mismatch in the stacking protocol configured. Some configuration errors can produce the No Neighbor state at one end and the Link Down state at the other end. Check the configuration at each port and correct as necessary.

  • If the show stacking command displays the node status as Disabled, you need to enable stacking on that node with the enable stacking command. You can enable stacking on the node from the master node, and you can reboot the disabled node from the master node to activate the slot number configuration.

  • If the show stacking command shows the stack state for a slot as Failed, check the following:
    • Does the show stacking stack-ports command show a port state as Inhibited? If so, the problem might be a duplicate slot number. If more than one node is using the same slot number, change the slot number on one of the affected nodes to a unique slot number.

    • Is the affected node isolated by other nodes for which the stack state is listed as Disabled? If so, you need to enable stacking on the disabled nodes.

    • Enter the show slot detail command. If the command displays License Mismatch, either upgrade the node license, or configure a license level restriction so that all master-capable nodes are at the same effective license level.

    • Enter the show slot detail command. If the command displays Incompatible EXOS Version, log into the master node and use the synchronize slot command to update the failed node.

  • If the show stacking command shows the stack state for a slot as Active and the “O” flag is set, check to see if the node is isolated from other stack nodes by a failed node.