You can use the upgrade process to
replace a faulty node in a multi-node TPVM deployment.
Before you begin
Ensure the cluster with the faulty node is running EFA 2.3.0 or later.
Ensure you have completed the high-availability prerequisites in EFA Requirements.
Ensure that EFA is not deployed on the
replacement node.
About this task
During this process, the faulty node is decommissioned, the replacement node is provisioned,
and the active node is reconfigured to form the cluster.
Perform this procedure on the active node where EFA is installed.
Procedure
-
Enter SLX Linux mode and copy
the EFA tar file to the SLX device.
device# start-shell
scp <username>@<hostip>:<buildpath>/efa-3.1.0.tar.gz
-
Deploy EFA on TPVM from the SLX shell.
device# efa deploy
Starting "efa deploy", DO NOT hit CTRL+C
Step 1: Checking if TPVM is deployed ...
Step 2: Get IP Addressed assigned to TPVM to deploy EFA
IP Address of the TPVM 10.x.x.x
Step 3: Checking for EFA packages in /efaboot directory
Step 4: Deploying EFA package efa-2.x.x.tar.gz on 10.x.x.x
The EFA Installer begins in a series of dialogs.
-
When prompted, select Multi Node Build Upgrade with
Node Replacement and OK.

Tip
Use arrow keys to move between options and the space bar to
select an option.
-
When prompted, enter the IP
address or host name of the replacement peer node and select
OK.
The node replacement
proceeds. Messages indicate the progress and when the replacement is
complete.
-
Verify the status of EFA after the node replacement.
device# sudo efactl status