You can set custom hostname for control nodes on Cloud Pak for Data System. If NPS is installed, additional steps might be
required.
Before you begin
Ensure that you have:
- Management network IPs for control nodes: node1, node2, node3
Procedure
- Connect to mgmt1 IP of e1n1.
- For management network IPs, update the IP values or custom hostnames, or both by
following Node specific information.
- Validate the YAML file.
- Test the YAML file and run the
playbooks.
- Validate that all nodes are accessible through their management IP addresses
(
nodes 1 - 3
and floating_ip
).
- If you have NPS installed and the Cloud Pak for Data System version is 1.x, follow these steps to reflect the
change on NPS containers:
- Upgrade the container, following the steps as described in Upgrading the NPS container. The upgrade must be done to the
same version as the one currently installed. By doing so, you ensure that the changes are applied to
all control nodes. To check the current container version,
run:
docker inspect ipshost1 | grep -i version
- After the container upgrade is completed, do not start NPS. Keep NPS in the Stopped state.
- Redeploy the container on other nodes as described in Redeploying the container
- When finished, start NPS: as the
nz
user, inside the container, run
nzstart
.
Note: After changing the hostnames the NPS container will rely on the external DNS server to resolve
the names of bare metal hosts. If you are using NPS, new hostnames have to be resolvable by external
DNS. If this condition is not met the NPS will fail to start with message ERROR: There should
be exactly one ip address found for <new hostname>.
On Cloud Pak for Data System version 2.x no additional steps are required as the pod with NPS host runs on a worker node.