PowerFlex PXE boot failing on 14G servers with X710 NICs

• OS installation is timing out during deployment.
• While the node is doing the PXE boot, you will see "Connection timed out" messages on iDRAC console.

• Error message in the rackserver-<service tag>_exception.log file:

#<ASM::UserException: The installation of the operating system for node 3Z5ZDF3 10.5.42.165 timed out.>

/opt/asm-deployer/lib/asm/razor.rb:314:in `block_until_task_complete'
/opt/asm-deployer/lib/asm/type/server.rb:3428:in `razor_block_until'
/opt/asm-deployer/lib/asm/type/server.rb:1687:in `provision_server!'
/opt/asm-deployer/lib/asm/service_deployment.rb:5146:in `process_server_with_types'
/opt/asm-deployer/lib/asm/service_deployment.rb:2949:in `process_server'
/opt/asm-deployer/lib/asm/service_deployment.rb:1409:in `block in create_component_thread'

• This is a regression from a change to add the LACP port channel (with fallback enabled) to the PXE switch interfaces before the PXE install process.
• Prior to this change, trunk ports were used. The new method works fine on Mellanox NICs but seems to be failing on Intel X710 NICs.

• When the node begins the PXE boot, you see Connection timed out messages on the iDRAC console.
• Now remove the port channel from the PXE switch interface.
  • Identify the ports used for the PXE network for each node in the deployment and the find the channel group # configured on each port (example below):

• Enter config mode on the interface for the first node and run the following command:

no channel-group <channel-group #> mode active

• Repeat the above steps for each node in the service.
• Save your config changes on the switch.

copy run start

• When the node reboots (should be in a boot loop due to PXE boot failure), the PXE install will go through.
• If the service has already failed, you will have to make the port channel changes and retry the service.
• Reboot the nodes and ensure that ESX is completely installed using PXE before retrying the service.