Restoring After a Failed Upgrade

17.2.12 Restoring After a Failed Upgrade

If the upgrade fails at any point, you can restore back to the Oracle Access Management (OAM) 12c deployment using the following steps:

Shut down the OAM 14c deployment using the following command:

 kubectl patch domain <domain> -n <domain_namespace> --type=merge --patch "{\"spec\": {\"serverStartPolicy\": \"Never\"}}"

For example:

 kubectl patch domain accessdomain -n oamns --type=merge --patch "{\"spec\": {\"serverStartPolicy\": \"Never\"}}"

The output will look similar to the following:

domain.weblogic.oracle/accessdomain patched

Run the following kubectl command to view the pods:

kubectl get pods -n <domain_namespace>

For example:

kubectl get pods -n oamns

The output will look similar to the following:


NAME                                                     READY   STATUS        RESTARTS   AGE
accessdomain-adminserver                                 1/1     Running       0          3h52m
accessdomain-oam-policy-mgr1                             1/1     Running       0          3h44m
accessdomain-oam-server1                                 1/1     Terminating   0          3h44m

The Administration Server pods and Managed Server pods will move to a STATUS of Terminating.

After a few minutes, run the command again and make sure the pods should have disappeared before continuing.

Restore the persistent volume from the backup taken before the upgrade:

sudo cp -rp <persistent_volume>/accessdomainpv <persistent_volume>/accessdomain_bkp14c

sudo rm -rf <persistent_volume>/accessdomainpv

sudo cp -rp <persistent_volume>/accessdomainpv_bkp12c <persistent_volume>/accessdomain

For example:

sudo cp -rp /nfs_volumes/oam/accessdomainpv /nfs_volumes/oam/accessdomain_bkp14c

sudo rm -rf /nfs_volumes/oam/accessdomainpv

sudo cp -rp /nfs_volumes/oam/accessdomainpv_bkp12c /nfs_volumes/oam/accessdomain

Restore the Oracle Database from the backup taken before the upgrade.
Run the following command to update the deployment with the OAM 12c container image used previously:
```
kubectl patch domain %DOMAIN_UID% -n %NAMESPACE% --type merge  -p '{"spec":{"image":"%WEBLOGIC_IMAGE%"}}'
```
For example:
- If using Oracle Container Registry or your own container registry for your OAM container image:
```
kubectl patch domain accessdomain -n oamns \
--type merge  -p '{"spec":{"image":"container-registry.oracle.com/middleware/oam_cpu:12.2.1.4-jdk8-ol8-<YYMMDD>"}}'
```
- If you are not using a container registry and have loaded the image on each of the worker nodes:
```
kubectl patch domain accessdomain -n oamns \
--type merge  -p '{"spec":{"image":"oracle/oam:12.2.1.4.0"}}'
```
  The output will look similar to the following:
```
domain.weblogic.oracle/accessdomain patched
```
Downgrade the WebLogic Kubernetes Operator to a version supported by OAM 12c. Follow the instructions at, Updating the WebLogic Kubernetes Operator, but use a supported operator for OAM 12c.

Run the following command to start the OAM domain:

kubectl patch domain.v9.weblogic.oracle "%DOMAIN_UID%" -n "%NAMESPACE%" \
--type=merge --patch "{\"spec\": {\"serverStartPolicy\": \"IfNeeded\"}}"

For example:

kubectl patch domain.v9.weblogic.oracle "accessdomain" -n "oamns" \
--type=merge --patch "{\"spec\": {\"serverStartPolicy\": \"IfNeeded\"}}"

The output will look similar to the following:

domain.weblogic.oracle/accessdomain patched

Run the following command to view the pods:

kubectl get pods -n %NAMESPACE%

For example:

kubectl get pods -n oamns

The output will look similar to the following:

NAME                                                     READY   STATUS        RESTARTS   AGE
accessdomain-introspector-jwqxw                          1/1     Running     0          10s

The introspect job will start, followed by the Administration Server pod, and then the OAM server pods. This process will take several minutes, so keep executing the command until all the pods are running with READY status 1/1:

Note:

Alternatively, you can add the watch flag, -w, which allows you watch the status of the pods as they change.

NAME                                                     READY   STATUS      RESTARTS   AGE  
accessdomain-adminserver                                 1/1     Running     0          10m
accessdomain-oam-policy-mgr1                             1/1     Running     0          7m35s
accessdomain-oam-server1                                 1/1     Running     0          7m35s

Once everything is running, check the consoles are accessible as per Validating the Domain URLs.