You can use the upgrade process to
replace a faulty node in a multi-node deployment.
Before you begin
- Ensure the cluster with the faulty node is
running EFA 2.5.5 or later.
- Ensure you have completed the
high-availability prerequisites in High-Availability Requirements.
- Ensure that XCO is not deployed on the
replacement node.
- Ensure that the faulty node is shutdown.
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 XCO is installed.
Procedure
-
Navigate to the directory where
the XCO file (*.tar.gz) is untarred.
-
Run the deployment script.
$source deployment.sh -I --deploy-suite fabric --deploy-type multi-node --deployipmode ipv4 --virtual-ipv4 10.32.85.119 --replacement-ip 10.20.48.101
Checking for EFA Stack...
Deployment mode is upgrade
Verifying connectivity to 10.32.85.114...
10.20.48.101 server is reachable...
You have entered:
- to redeploy EFA at version 3.3.0 build 12
- with peer 10.20.48.101
- and with VIP 10.32.85.119
- with node replacement
- with IP Stack ipv4
- with suites: Fabric Automation
Verifying if monitor service is running on 10.32.85.111 10.20.48.101...
Checking system configuration on 10.32.85.111 10.20.48.101...
Ensuring machine clocks are in sync
Verifying clocks are approximately in sync
Checking default gateway reachability on all nodes...
Completed verification of default gateway reachability on all nodes
Ensuring peer hostnames are unique
Verifying unique hostname between nodes
Hostnames are unique
Ensuring compatible OS version
Verifying Operating System between nodes
Operating system of all nodes are same
Making backup
Removing legacy EFA installation
Stopping EFA services
Undeploying EFA application...
Undeploying ecosystem services
Undeploying core services
Removed current application deployment successfully.
Removing EFA container images
Removing container images on 10.32.85.111 10.20.48.101...
Removing EFA OS services
Removing k3s container orchestration
Removing database
Unholding mariadb server
Removing Database Server
Unholding mariadb client
Removing Database Client
Removing cluster filesystem
Unholding glusterfs
Removing keepalived for cluster virtual IP
Removing database sync tools
Removing EFA services and utilities
Proceeding with new EFA installation
Verifying system requirements
Verifying system requirements on all nodes......
The node replacement proceeds. Messages indicate the progress and when the
replacement is complete.
-
Verify the status of XCO after the node replacement.
Ensure that status of all the nodes are up.
