Upgrade your Red Hat®
OpenShift® Container Platform when the version is no longer
supported by Cloud Pak for Business Automation.
Before you begin
Before you start the upgrade of Red Hat
OpenShift Container Platform, make sure that you take a
backup so that you can recover from any upgrade issues. For more information, see Backing up
your environments.
The Red Hat
OpenShift Container Platform version 4.12
is the minimum required version for Cloud Pak for Business Automation
24.0.0. If the
Red Hat
OpenShift Container Platform version is
4.12.xx, the xx is the security or bug fix part of the
version. It is recommended to apply the latest security or bug fixes for a specific Red Hat
OpenShift Container Platform minor version.
Attention: Before you begin the upgrade of the OpenShift cluster, make sure that all the
standby
EDB Postgres cluster
members are up and running.
Under normal circumstances, an EDB Postgres cluster can manage if one
member is down, but it is not ideal just before you start an OpenShift upgrade. The status of the
EDB cluster can be checked in the EDB Postgres CR.
For more information
about EDB Postgres maintenance on
Kubernetes, see the EDB Postgres
documentation.
About this task
When you need to upgrade your OpenShift Container Platform, make sure that you
refer to the steps that are needed for the cluster that contains Cloud Pak for Business Automation.
Note: The previous Red Hat OpenShift upgrade process transitions through all the Red Hat OpenShift
versions. For example, if you needed to upgrade Red Hat OpenShift from 4.12 to 4.14, then you need
to upgrade to each Red Hat OpenShift version step by step (4.12 -> 4.13 -> 4.14). However, Red
Hat now supports an Extended Update Support (EUS) upgrade from 4.12 -> 4.14, see
Red Hat OpenShift EUS Overview.
Each upgrade can take a few hours, so plan for the entire
upgrade. The expectation is that if you upgrade to an intermediate version (odd versions), and you
complete the upgrade within a reasonable time frame. Ideally, the cluster does not run on an
intermediate version indefinitely.
The Red Hat OpenShift Container Platform Update Graph has a tool that provides the appropriate
minor versions to use for the intermediate updates. Click the Update Path
tile to access the tool, and you do need Red Hat customer account credentials to use it.
Procedure
- Go to IBM Cloud Pak® for Business Automation detailed system
requirements page and select your Cloud Pak for Business Automation version. Click the
Containers tab on this report to see the supported minor OpenShift Container
Platform releases for the given Cloud Pak for Business Automation version. It also
includes any Cloud Pak for Business Automation interim fix requirements for a specific OpenShift Container Platform version.
Note: If your current OpenShift Container Platform version is too old for your Cloud Pak for Business Automation version, then
upgrade your OpenShift Container Platform before you upgrade Cloud Pak for Business Automation. If your Cloud Pak for Business Automation installation is too
old for the OpenShift Container Platform version, then upgrade Cloud Pak for Business Automation.
- Various Postgres deployments have a
PodDisruptionBudget
(PDB) with
minimum available
set to 1. If the replica size is also set to 1, then it can
prevent nodes from restarting automatically. If a node restart is blocked, it can prevent the
completion of the OpenShift Container Platform upgrade or other cluster configuration tasks.
To mitigate the problem, scale down the Postgres operator or delete the pods that are stuck.
- Postgres deployments that have a PodDisruptionBudget (PDB).
- Scale down the
postgresql-operator
deployment to replica 0.
- Delete the PDB objects from the CP4BA namespace with min available set to 1. The PDB objects
must be related to deployments owned by the cluster CR objects.
- Complete the OpenShift Container Platform upgrade.
- Scale the
postgresql-operator
operator back up to replica 1 after the
upgrade.
- When the nodes attempt to restart during the upgrade or configuration task, delete the pods from
the Cloud Pak for Business Automation
namespace that are stuck. The pods must be related to deployments owned by the cluster CR
objects.