In this section you learn how to upgrade OUD from a previous version. Follow the section relevant to the version you are upgrading from.
The instructions below are for upgrading from October 22 (22.4.1) to January 23 (22.1.1) or later.
Note: If you are not using Oracle Container Registry or your own container registry, then you must first load the new container image on all nodes in your Kubernetes cluster.
You can update the deployment with a new OUD container image using one of the following methods:
Navigate to the $WORKDIR/kubernetes/helm directory:
$ cd $WORKDIR/kubernetes/helm
Create a oud-patch-override.yaml file that contains:
image:
repository: <image_location>
tag: <image_tag>
imagePullSecrets:
- name: orclcred
For example:
image:
repository: container-registry.oracle.com/middleware/oud_cpu
tag: 12.2.1.4-jdk8-ol7-new
imagePullSecrets:
- name: orclcred
The following caveats exist:
If you are not using Oracle Container Registry or your own container registry for your OUD container image, then you can remove the following:
imagePullSecrets:
- name: orclcred
Run the following command to upgrade the deployment:
$ helm upgrade --namespace <namespace> \
--values oud-patch-override.yaml \
<release_name> oud-ds-rs --reuse-values
For example:
$ helm upgrade --namespace oudns \
--values oud-patch-override.yaml \
oud-ds-rs oud-ds-rs --reuse-values
--set argumentNavigate to the $WORKDIR/kubernetes/helm directory:
$ cd $WORKDIR/kubernetes/helm
Run the following command to update the deployment with a new OUD container image:
$ helm upgrade --namespace <namespace> \
--set image.repository=<image_location>,image.tag=<image_tag> \
--set imagePullSecrets[0].name="orclcred" \
<release_name> oud-ds-rs --reuse-values
For example:
$ helm upgrade --namespace oudns \
--set image.repository=container-registry.oracle.com/middleware/oud_cpu,image.tag=12.2.1.4-jdk8-ol7-new \
--set imagePullSecrets[0].name="orclcred" \
oud-ds-rs oud-ds-rs --reuse-values
The following caveats exist:
--set imagePullSecrets[0].name="orclcred".After updating with the new image the pods will restart. Verify the pods are running:
$ kubectl --namespace <namespace> get pods
For example:
$ kubectl --namespace oudns get pods
The output will look similar to the following:
NAME READY STATUS RESTARTS AGE IP NODE NOMINATED NODE READINESS GATES
pod/oud-ds-rs-0 1/1 Running 0 45m 10.244.0.195 <Worker Node> <none> <none>
pod/oud-ds-rs-1 1/1 Running 0 45m 10.244.0.194 <Worker Node> <none> <none>
pod/oud-ds-rs-2 1/1 Running 0 45m 10.244.0.193 <Worker Node> <none> <none>
Note: It will take several minutes before the pods start. While the oudsm pods have a STATUS of 0/1 the pods are started but the OUD server associated with it is currently starting.
Verify the pods are using the new image by running the following command:
$ kubectl describe pod <pod> -n <namespace>
For example:
$ kubectl describe pod oud-ds-rs-0 -n oudns
The output will look similar to the following:
Name: oud-ds-rs-0
Namespace: oudns
Priority: 0
Node: <Worker Node>/100.102.48.28
Start Time: <DATE>
Labels: app.kubernetes.io/instance=oud-ds-rs
app.kubernetes.io/managed-by=Helm
app.kubernetes.io/name=oud-ds-rs
app.kubernetes.io/version=12.2.1.4.0
helm.sh/chart=oud-ds-rs-0.1
oud/instance=oud-ds-rs-0
Annotations: meta.helm.sh/release-name: oud-ds-rs
meta.helm.sh/release-namespace: oudns
Status: Running
IP: 10.244.1.44
etc...
Events:
Type Reason Age From Message
---- ------ ---- ---- -------
Normal Killing 4m26s kubelet Container oud-ds-rs definition changed, will be restarted
Warning Unhealthy 3m56s kubelet Readiness probe failed:
Normal Pulling 3m56s kubelet Pulling image "container-registry.oracle.com/middleware/oud_cpu:12.2.1.4-jdk8-ol7-new"
Warning Unhealthy 3m27s kubelet Liveness probe failed: dial tcp 10.244.1.44:1389: connect: connection refused
Normal Created 3m22s (x2 over 142m) kubelet Created container oud-ds-rs
Normal Started 3m22s (x2 over 142m) kubelet Started container oud-ds-rs
Normal Pulled 3m22s kubelet Successfully pulled image "container-registry.oracle.com/middleware/oud_cpu:12.2.1.4-jdk8-ol7-new" in 33.477063844s
The instructions below are for upgrading from Jul 22 (22.3.1) to January 23 (23.1.1) or later.
Note: If you are not using Oracle Container Registry or your own container registry, then you must first load the new container image on all nodes in your Kubernetes cluster.
You can update the deployment with a new OUD container image using one of the following methods:
Navigate to the $WORKDIR/kubernetes/helm directory:
$ cd $WORKDIR/kubernetes/helm
Create a oud-patch-override.yaml file that contains:
image:
repository: <image_location>
tag: <image_tag>
imagePullSecrets:
- name: orclcred
For example:
image:
repository: container-registry.oracle.com/middleware/oud_cpu
tag: 12.2.1.4-jdk8-ol7-new
imagePullSecrets:
- name: orclcred
The following caveats exist:
If you are not using Oracle Container Registry or your own container registry for your OUD container image, then you can remove the following:
imagePullSecrets:
- name: orclcred
Run the following command to upgrade the deployment:
$ helm upgrade --namespace <namespace> \
--values oud-patch-override.yaml \
<release_name> oud-ds-rs --reuse-values
For example:
$ helm upgrade --namespace oudns \
--values oud-patch-override.yaml \
oud-ds-rs oud-ds-rs --reuse-values
--set argumentNavigate to the $WORKDIR/kubernetes/helm directory:
$ cd $WORKDIR/kubernetes/helm
Run the following command to update the deployment with a new OUD container image:
$ helm upgrade --namespace <namespace> \
--set image.repository=<image_location>,image.tag=<image_tag> \
--set imagePullSecrets[0].name="orclcred" \
<release_name> oud-ds-rs --reuse-values
For example:
$ helm upgrade --namespace oudns \
--set image.repository=container-registry.oracle.com/middleware/oud_cpu,image.tag=12.2.1.4-jdk8-ol7-new \
--set imagePullSecrets[0].name="orclcred" \
oud-ds-rs oud-ds-rs --reuse-values
The following caveats exist:
--set imagePullSecrets[0].name="orclcred".After updating with the new image the pods will restart. Verify the pods are running:
$ kubectl --namespace <namespace> get pods
For example:
$ kubectl --namespace oudns get pods
The output will look similar to the following:
NAME READY STATUS RESTARTS AGE IP NODE NOMINATED NODE READINESS GATES
pod/oud-ds-rs-0 1/1 Running 0 45m 10.244.0.195 <Worker Node> <none> <none>
pod/oud-ds-rs-1 1/1 Running 0 45m 10.244.0.194 <Worker Node> <none> <none>
pod/oud-ds-rs-2 1/1 Running 0 45m 10.244.0.193 <Worker Node> <none> <none>
Note: It will take several minutes before the pods start. While the oudsm pods have a STATUS of 0/1 the pods are started but the OUD server associated with it is currently starting.
Verify the pods are using the new image by running the following command:
$ kubectl describe pod <pod> -n <namespace>
For example:
$ kubectl describe pod oud-ds-rs-0 -n oudns
The output will look similar to the following:
Name: oud-ds-rs-0
Namespace: oudns
Priority: 0
Node: <Worker Node>/100.102.48.28
Start Time: <DATE>
Labels: app.kubernetes.io/instance=oud-ds-rs
app.kubernetes.io/managed-by=Helm
app.kubernetes.io/name=oud-ds-rs
app.kubernetes.io/version=12.2.1.4.0
helm.sh/chart=oud-ds-rs-0.1
oud/instance=oud-ds-rs-0
Annotations: meta.helm.sh/release-name: oud-ds-rs
meta.helm.sh/release-namespace: oudns
Status: Running
IP: 10.244.1.44
etc...
Events:
Type Reason Age From Message
---- ------ ---- ---- -------
Normal Killing 4m26s kubelet Container oud-ds-rs definition changed, will be restarted
Warning Unhealthy 3m56s kubelet Readiness probe failed:
Normal Pulling 3m56s kubelet Pulling image "container-registry.oracle.com/middleware/oud_cpu:12.2.1.4-jdk8-ol7-new"
Warning Unhealthy 3m27s kubelet Liveness probe failed: dial tcp 10.244.1.44:1389: connect: connection refused
Normal Created 3m22s (x2 over 142m) kubelet Created container oud-ds-rs
Normal Started 3m22s (x2 over 142m) kubelet Started container oud-ds-rs
Normal Pulled 3m22s kubelet Successfully pulled image "container-registry.oracle.com/middleware/oud_cpu:12.2.1.4-jdk8-ol7-new" in 33.477063844s
Upgrade Elasticsearch and Kibana by following Upgrading Elasticsearch and Kibana.
In releases prior to July 22 (22.3.1) OUD used pod based deployment. From July 22 (22.3.1) onwards OUD is deployed using StatefulSets.
If you are upgrading from a release prior to July 22 (22.3.1) you must follow the steps below to deploy a new OUD instance to use your existing OUD data in <persistent_volume>/oud_user_projects.
Note: The steps below will incur a small outage.
Find the deployment release name as follows:
$ helm --namespace <namespace> list
For example:
$ helm --namespace oudns list
The output will look similar to the following:
NAME NAMESPACE REVISION UPDATED STATUS CHART APP VERSION
oud-ds-rs oudns 1 2022-07-11 09:46:17.613632382 +0000 UTC deployed oud-ds-rs-0.2 12.2.1.4.0
Delete the deployment using the following command:
$ helm uninstall --namespace <namespace> <release>
For example:
$ helm uninstall --namespace oudns oud-ds-rs
release "oud-ds-rs" uninstalled
Run the following command to view the status:
$ kubectl --namespace oudns get pod,service,secret,pv,pvc,ingress -o wide
Initially the pods and persistent volume (PV) and persistent volume claim (PVC) will move to a Terminating status:
NAME READY STATUS RESTARTS AGE IP NODE NOMINATED NODE READINESS GATES
pod/oud-ds-rs-0 1/1 Terminating 0 24m 10.244.1.180 <Worker Node> <none> <none>
pod/oud-ds-rs-1 1/1 Terminating 0 18m 10.244.1.181 <Worker Node> <none> <none>
pod/oud-ds-rs-2 1/1 Terminating 0 12m 10.244.1.182 <Worker Node> <none> <none>
NAME TYPE DATA AGE
secret/default-token-msmmd kubernetes.io/service-account-token 3 3d20h
secret/dockercred kubernetes.io/dockerconfigjson 1 3d20h
secret/orclcred kubernetes.io/dockerconfigjson 1 3d20h
NAME CAPACITY ACCESS MODES RECLAIM POLICY STATUS CLAIM STORAGECLASS REASON AGE VOLUMEMODE
persistentvolume/oud-ds-rs-pv 20Gi RWX Delete Terminating oudns/oud-ds-rs-pvc manual 24m Filesystem
NAME STATUS VOLUME CAPACITY ACCESS MODES STORAGECLASS AGE VOLUMEMODE
persistentvolumeclaim/oud-ds-rs-pvc Terminating oud-ds-rs-pv 20Gi RWX manual 24m Filesystem
Run the command again until the pods, PV and PVC disappear.
Create a working directory on the persistent volume to setup the latest source code:
$ mkdir <persistent_volume>/<workdir>
For example:
$ mkdir /scratch/shared/OUDK8Slatest
Download the latest OUD deployment scripts from the OUD repository:
$ cd <persistent_volume>/<workdir>
$ git clone https://github.com/oracle/fmw-kubernetes.git --branch release/23.1.1
For example:
$ cd /scratch/shared/OUDLatestSource
$ git clone https://github.com/oracle/fmw-kubernetes.git --branch release/23.1.1
Set the $WORKDIR environment variable as follows:
$ export WORKDIR=<workdir>/fmw-kubernetes/OracleUnifiedDirectory
For example:
$ export WORKDIR=/scratch/shared/OUDK8Slatest/fmw-kubernetes/OracleUnifiedDirectory
Navigate to the $WORKDIR/kubernetes/helm directory
$ cd $WORKDIR/kubernetes/helm
Create an oud-ds-rs-values-override.yaml as follows:
image:
repository: <image_location>
tag: <image_tag>
pullPolicy: IfNotPresent
imagePullSecrets:
- name: orclcred
oudConfig:
rootUserPassword: <password>
persistence:
type: filesystem
filesystem:
hostPath:
path: <persistent_volume>/oud_user_projects
cronJob:
kubectlImage:
repository: bitnami/kubectl
tag: <version>
pullPolicy: IfNotPresent
imagePullSecrets:
- name: dockercred
For example:
image:
repository: container-registry.oracle.com/middleware/oud_cpu
tag: 12.2.1.4-jdk8-ol7-<jul22>
pullPolicy: IfNotPresent
imagePullSecrets:
- name: orclcred
oudConfig:
rootUserPassword: <password>
persistence:
type: filesystem
filesystem:
hostPath:
path: /scratch/shared/oud_user_projects
cronJob:
kubectlImage:
repository: bitnami/kubectl
tag: 1.21.6
pullPolicy: IfNotPresent
imagePullSecrets:
- name: dockercred
The following caveats exist:
The <persistent_volume>/oud_user_projects must point to the directory used in your previous deployment otherwise your existing OUD data will not be used. Make sure you take a backup of the <persistent_volume>/oud_user_projects directory before proceeding further.
Replace <password> with the password used in your previous deployment.
The <version> in kubectlImage tag: should be set to the same version as your Kubernetes version (kubectl version). For example if your Kubernetes version is 1.21.6 set to 1.21.6.
If you are not using Oracle Container Registry or your own container registry for your OUD container image, then you can remove the following:
imagePullSecrets:
- name: orclcred
If using NFS for your persistent volume then change the persistence section as follows:
persistence:
type: networkstorage
networkstorage:
nfs:
path: <persistent_volume>/oud_user_projects
server: <NFS IP address>
Run the following command to deploy OUD:
$ helm install --namespace <namespace> \
--values oud-ds-rs-values-override.yaml \
<release_name> oud-ds-rs
For example:
$ helm install --namespace oudns \
--values oud-ds-rs-values-override.yaml \
oud-ds-rs oud-ds-rs
Check the OUD deployment as per Verify the OUD deployment and Verify the OUD replication.
Upgrade Elasticsearch and Kibana by following Upgrading Elasticsearch and Kibana.
This section shows how to upgrade Elasticsearch and Kibana. From October 22 (22.4.1) onwards, OUD logs should be stored on a centralized Elasticsearch and Kibana stack.
Note: This section should only be followed if upgrading from July 22 (22.3.1) or earlier to January 23 (23.1.1). If you are upgrading from October 22 to January 23 do not follow this section.
If you haven’t already downloaded the latest code repository, do so as follows:
Create a working directory to setup the source code.
$ mkdir <workdir>
For example:
$ mkdir /scratch/shared/OUDK8Slatest
Download the latest OUD deployment scripts from the OUD repository.
$ cd <workdir>
$ git clone https://github.com/oracle/fmw-kubernetes.git --branch release/23.1.1
For example:
$ cd /scratch/OUDK8Slatest
$ git clone https://github.com/oracle/fmw-kubernetes.git --branch release/23.1.1
Set the $WORKDIR environment variable as follows:
$ export WORKDIR=<workdir>/fmw-kubernetes/OracleUnifiedDirectory
For example:
$ export WORKDIR=/scratch/shared/OUDK8Slatest/fmw-kubernetes/OracleUnifiedDirectory
From October 22 (22.4.1) onwards, OUD logs should be stored on a centralized Elasticsearch and Kibana (ELK) stack.
Deployments prior to October 22 (22.4.1) used local deployments of Elasticsearch and Kibana.
If you are upgrading from July 22 (22.3.1) or earlier, to January 23 (23.1.1) or later, you must first undeploy Elasticsearch and Kibana using the steps below:
Navigate to the $WORKDIR/kubernetes/helm directory and create a logging-override-values-uninstall.yaml with the following:
elk:
enabled: false
Run the following command to remove the existing ELK deployment:
$ helm upgrade --namespace <domain_namespace> --values <valuesfile.yaml> <releasename> oud-ds-rs --reuse-values
For example:
$ helm upgrade --namespace oudns --values logging-override-values-uninstall.yaml oud-ds-rs oud-ds-rs --reuse-values