Beginning with OME-M 1.20.00 and OS10.5.0.7, the ability to rebalance FC and FCoE sessions across FC uplinks has been added. This can be validated in Scenario 5: Connect MX9116n FSE to Fibre Channel storage - NPIV Proxy Gateway mode.
The system performs an end-node based rebalancing when the CLI command is run. Factors for rebalancing are the current session count on the uplink includes Fabric Login requests (FLOGI), Fabric Discovery Requests (FDISC), and the speed of the uplink. Rebalance can be applied once the FC fabric is up and running and uplinks sessions are established in them.
Prior to the release of Dell SmartFabric OS10.5.1, NPG implementations exposed one Fibre Channel Forwarder (FCF) for each physical FC uplink to end nodes. Starting with Dell SmartFabric OS10.5.2.4, all physical uplink interfaces within a vFabric are represented as a single logical FCF. This improves session management and failover as the CNA no longer has to select a different FCF during a link event.
Requirements and configuration guidelines
When a new physical uplink is added to a vFabric operating in NPG mode, or when a physical uplink having FC/FCoE sessions established in them goes down, the system will go to an unbalanced state. A manual rebalance can be performed when the system is found to be unbalanced.
The new uplink added must be operationally up before the rebalance is triggered. When an uplink goes down, all the sessions associated with that uplink will be interrupted and will be reestablished and load balanced to the other available uplinks. Rebalancing is done at the vFabric level.
Because FC session rebalancing is path disruptive, the command provides the ability to perform a dry run to provide a list of servers that will be affected.
Below are the steps to perform rebalancing of uplinks.
System in unbalanced state
The following command shows that the system is in an unbalanced state. Run the show npg device brief
and show npg uplink-interfaces
commands to see the unbalanced state of the system. In the following figure, interface Fc 1/1/23
has two FC sessions and interface Fc 1/1/24
has zero.
Perform a dry run of the rebalance command
To understand what changes will be made during a rebalance before making them, run the re-balance npg sessions vfabric 10 dry-run
command to view what changes will be made when the command is executed.
Run rebalance command
To perform the rebalance, run the command re-balance npg sessions vfabric 10
.
System in a balanced state
The following figure shows the system in a balanced state. Interface Fc 1/1/23
now has one FC session and interface Fc 1/1/24
has one.
Beginning with Dell SmartFabric OS10.5.2, the show npg uplink-interface command has one more option added as fcf-info to display the FCF Availability Status, fabric name of the FC upstream switch connected, error reason, FCF advertisement delay timeout left, and duplicate FC id assignment counter.
MX9116N-A1# show npg uplink-interfaces fcf-info
Vfabric-Id : 10
FAD Timeout Left : 0 second(s)
FCF Availability Status : Yes
Uplink Duplicate
Intf Upstream Fabric-name Error Reason FC Id(s)
--------------------------------------------------------------------------
Fc 1/1/24 10:00:14:18:77:20:7f:cf NONE 0
Fc 1/1/23 10:00:14:18:77:20:7f:cf NONE 0