Prerequisites

Prerequisites for IBM Fusion HCI System upgrade procedure. Go through this topic for better planning.

General

You must be on IBM Fusion HCI System version 2.8.2 to move to 2.9.0.
IBM Fusion HCI System 2.9.0 supports OpenShift® Container Platform 4.12, 4.14, 4.15, and 4.16.
Note: For the procedure to upgrade OpenShift Container Platform, see OpenShift Container Platform upgrade . For storage support related information, see Data services version support matrix.
If you installed IBM Fusion HCI System in the previous version by using offline or online installation mode, then do not change the mode during the upgrade to the new version. To change the installation mode, reinstall IBM Fusion HCI System.
Logs that you collected by using the IBM Fusion Collect logs user interface page gets deleted after the upgrade process completes. Download the needed logs before you begin the upgrade. In addition, check system health from the log collections.
The IBM Fusion HCI System operator upgrade can get stuck during upgrade due to a known Red Hat OLM issue. The issue is specific to Red Hat OpenShift Container Platform versions 4.14, 4.15, and 4.16. Resolve the problem before you proceed with the upgrade. For the resolution steps, see Hot fix page.

Setup enterprise registry

If you installed the earlier version of IBM Fusion HCI System by using your enterprise registry, then follow the steps to mirror images in your enterprise registry.

Mirror IBM Fusion HCI System 2.9.0 images, IBM Storage Scale images, Fusion Data Foundation, Backup & Restore, and Data Cataloging images. For steps to mirror, see End-to-end mirroring images of IBM Fusion HCI System and its services to the enterprise registry.
Update the global pull secret with the mirror registry credentials to which the current version images are mirrored. If you are mirroring to the same enterprise registry that you used for the previous version, then ignore this step.

Modify the image content source policy isf-operator-index to add the new mirror that points to the new registry for each source defined in the image content source policy. If you are mirroring to the same enterprise registry that you used for the previous version, then ignore this step.

See the sample for image content source policy:

Note: After the IBM Fusion HCI System is upgraded, you can see all the new IBM Fusion services that are introduced in 2.9.0. If you want to install the new services, add image content source policy and the related image. For more information, see End-to-end mirroring images of IBM Fusion HCI System and its services to the enterprise registry.


apiVersion: operator.openshift.io/v1alpha1
kind: ImageContentSourcePolicy
metadata:
  name: isf-operator-index
spec:
  repositoryDigestMirrors:
  # for scale
  - mirrors:
    - "<Old ISF enterprise registry>/<Old ISF target-path>"
    - "<Old ISF enterprise registry>/<New ISF target-path>"
    source: cp.icr.io/cp/spectrum/scale
  - mirrors:
    - "<Old ISF enterprise registry>/<Old ISF target-path>"
    - "<Old ISF enterprise registry>/<New ISF target-path>"
    source: icr.io/cpopen
  #for IBM Spectrum Fusion operator 
  - mirrors:
    - "<old ISF enterprise registry>/<Old ISF target-path>"
    - "<Old ISF enterprise registry>/<New ISF target-path>"
    source: cp.icr.io/cp/isf
  
  # for ose-kube-rbac-proxy
  - mirrors:
    - "<New ISF enterprise registry>/<New ISF target-path>/openshift4/ose-kube-rbac-proxy"
    source: registry.redhat.io/openshift4/ose-kube-rbac-proxy

For offline, check whether the ibm-opencloud and ibm-operator-catalog catalog source are in an error state. If they are in an error state and not in use, then delete them.
- To validate if catalogsource are in use or not, run the following command to list all the subscriptions in IBM Fusion HCI System cluster.
```
oc get sub -A
```
  If the catalog source name under source fields does not show ibm-opencloud and ibm-operator-catalog in the output, then it is not in use.
- Run oc get catsrc -A command to ensure that the following catalog sources are present and in ready state:
  - isf-catalog or fusion-catalog
  - redhat-operators
- Run the following command to check whether olm pod catalog-operator is running in openshift-operator-lifecycle-management project:
```
oc -n openshift-operator-lifecycle-manager get po
```

Metro-DR

CSI configmap

Run the following command to tune CSI configmap:

oc -n ibm-spectrum-scale-csi create configmap ibm-spectrum-scale-csi-config --from-literal=VAR_DRIVER_DISCOVER_CG_FILESET=DISABLED

Prechecks-acks configmap
Follow the steps to update the prechecks-acks configmap:
1. Run the following command to check whether the configmap is present on the site 2 cluster.
```
oc get cm | grep precheck-acks
```
2. If the configmap is present on the cluster, then edit configmap using the following command and set the field GDPScaleUp to true.
```
oc edit cm precheck-acks
```
  For example:
```
data:
  GDPScaleUp: 'true'
```
3. If the configmap is not present on the cluster, then create a YAML file in the local system using following steps and apply it to the site 2 cluster.
  1. Run the following command to create a YAML file in the local system.
```
vi acks.yaml
```
  2. Add the following fields to the acks.yaml file.
```
apiVersion: v1
kind: ConfigMap
metadata:
  name: 'precheck-acks'
  namespace: ibm-spectrum-fusion-ns
data:
  GDPScaleUp: 'true'
```
  3. Run the following command to apply the acks.yaml to the site 2 cluster.
```
oc apply -f acks.yaml
```
    Sample output:
    oc apply -f acks.yaml configmap/precheck-acks created
  4. Run the following command to verify whether the configmap is created.
```
oc get cm precheck-acks -o yaml | grep GDPScaleUp
```
    Sample output:
    oc get cm precheck-acks -o yaml | grep GDPScaleUp GDPScaleUp: "true"

Network

If you are using the proxy in an online or offline environment, then make sure to add all the ipv6LLA of the rack nodes to the OpenShift Proxy. You can find the ipv6LLA values in the kickstart-config map in the IBM Fusion namespace.

The ipv6LLA values are added in the noProxy key inside the spec section.

oc edit proxy cluster

Example:


apiVersion: [config.openshift.io/v1](http://config.openshift.io/v1)
kind: Proxy
metadata:
  name: cluster
spec:
  httpProxy: http://<IP>:<PORT>/
  httpsProxy: http://<IP>:<PORT>/
  noProxy: .[base.test.net](http://base.test.net/),9999,metal3-state.openshift-machine-api,5050,6385,8089,6180,80,8084,fe80::72e2:84ff:fe23:78df,fd8c:215d:178e:c0de:72e2:84ff:fe23:78df,fe80::72e2:84ff:fe23:77f3,fd8c:215d:178e:c0de:72e2:84ff:fe23:77f3,fe80::72e2:84ff:fe23:78c3,fd8c:215d:178e:c0de:72e2:84ff:fe23:78c3

Node

Do not start node upsize or scale disk scale out until upgrade is complete.
Make sure that no operation is done on OpenShift Container Platform cluster that causes machine configuration rollout, for example:
- Node maintenance
- Node reboot
- Old firmware upgrade
- Image content source policy update
- Pull secret updates