Starting with the 2.2.0 release, the operator supports FMW Infrastructure domains, that is, domains that are created with the FMW Infrastructure installer rather than the WebLogic Server installer. These domains contain the Java Required Files (JRF) feature and are the prerequisite for “upper stack” products like Oracle SOA Suite, for example. These domains also require a database and the use of the Repository Creation Utility (RCU).
This document provides details about the special considerations for running FMW Infrastructure domains with the operator. Other than those considerations listed here, FMW Infrastructure domains work in the same way as WebLogic Server domains. The remainder of the documentation in this site applies equally to FMW Infrastructure domains and WebLogic Server domains.
FMW Infrastructure domains are supported using the Domain in PV domain home source type only.
For more information about the deployment of Oracle Fusion Middleware products on Kubernetes, see https://oracle.github.io/fmw-kubernetes/.
Compared to running a WebLogic Server domain in Kubernetes using the operator, the following limitations currently exist for FMW Infrastructure domains:
The WebLogic Kubernetes Operator requires patch 29135930.
The standard pre-built FMW Infrastructure General Availability image, container-registry.oracle.com/middleware/fmw-infrastructure:12.2.1.3, already
has this patch applied. The FMW Infrastructure 12.2.1.4.0 images do not require this patch. For detailed instructions
on how to log in to the Oracle Container Registry and accept license agreement, see this document.
NOTE: As of December, 2022, Fusion Middleware 12.2.1.3 is no longer supported. The last Critical Patch Updates (CPU) images for FMW Infrastructure 12.2.1.3 were published in October, 2022.
This sample uses General Availability (GA) images. GA images are suitable for demonstration and development purposes only where the environments are not available from the public Internet; they are not acceptable for production use. In production, you should always use CPU (patched) images from OCR or create your images using the WebLogic Image Tool (WIT) with the --recommendedPatches option. For more guidance, see Apply the Latest Patches and Updates in Securing a Production Environment for Oracle WebLogic Server.
To pull an image from the Oracle Container Registry, in a web browser, navigate to https://container-registry.oracle.com and log in using the Oracle Single Sign-On authentication service. If you do not already have SSO credentials, at the top of the page, click the Sign In link to create them.
Use the web interface to accept the Oracle Standard Terms and Restrictions for the Oracle software images that you intend to deploy. Your acceptance of these terms is stored in a database that links the software images to your Oracle Single Sign-On login credentials.
First, you will need to log in to the Oracle Container Registry:
$ docker login container-registry.oracle.com
Then, you can pull the image with this command:
$ docker pull container-registry.oracle.com/middleware/fmw-infrastructure:12.2.1.4
If desired, you can:
Check the WLS version with docker run container-registry.oracle.com/middleware/fmw-infrastructure:12.2.1.4 sh -c 'source $ORACLE_HOME/wlserver/server/bin/setWLSEnv.sh > /dev/null 2>&1 && java weblogic.version'
Check the WLS patches with docker run container-registry.oracle.com/middleware/fmw-infrastructure:12.2.1.4 sh -c '$ORACLE_HOME/OPatch/opatch lspatches'
Additional information about using this image is available in the Oracle Container Registry.
You can also create an image containing the FMW Infrastructure binaries. We provide a sample in the Oracle GitHub account that demonstrates how to create an image to run the FMW Infrastructure. Please consult the README file associated with this sample for important prerequisite steps, such as building or pulling the Server JRE image and downloading the Fusion Middleware Infrastructure installer binary.
After cloning the repository and downloading the installer from Oracle Technology Network or e-delivery, you create your image by running the provided script:
$ cd docker-images/OracleFMWInfrastructure/dockerfiles
$ ./buildDockerImage.sh -v 12.2.1.4 -s
The image produced will be named oracle/fmw-infrastructure:12.2.1.4.
You must also install the required patch to use this image with the operator. We provide a sample that demonstrates how to create an image with the necessary patch installed.
After downloading the patch from My Oracle Support, you create the patched image by running the provided script:
$ cd docker-images/OracleFMWInfrastructure/samples/12213-patch-fmw-for-k8s
$ ./build.sh
This will produce an image named oracle/fmw-infrastructure:12213-update-k8s.
All samples and instructions reference the pre-built image, container-registry.oracle.com/middleware/fmw_infrastructure:12.2.1.4. Because these samples build an image based on WebLogic Server 12.2.1.3 and use the tag, oracle/fmw-infrastructure:12213-update-k8s, be sure to update your sample inputs to use this image value.
These samples allow you to create an image containing the FMW Infrastructure binaries and the necessary patch. You can use this image to run the Repository Creation Utility and to run your domain using the “domain on a persistent volume” model. If you want to use the “domain in an image” model, you will need to go one step further and add another layer with your domain in it. You can use WLST or WDT to create your domain.
Before creating a domain, you will need to set up the necessary schemas in your database.
FMW Infrastructure domains require a database with the necessary schemas installed in them. We provide a utility, called the Repository Creation Utility (RCU), which allows you to create those schemas. You must set up the database before you create your domain. There are no additional requirements added by running FMW Infrastructure in Kubernetes; the same existing requirements apply.
For testing and development, you may choose to run your database inside Kubernetes or outside of Kubernetes.
The Oracle Database images are only supported for non-production use. For more details, see My Oracle Support note: Oracle Support for Database Running on Docker (Doc ID 2216342.1)
If you wish to run the database inside Kubernetes, you can use the official container image
from the Oracle Container Registry.
Please note that there is a Slim Variant (12.2.0.1-slim tag) of EE that has reduced
disk space (4GB) requirements and a quicker container startup.
Running the database inside the Kubernetes cluster is possibly more relevant or desirable in test/development or CI/CD scenarios.
Here is an example of a Kubernetes YAML file to define a deployment of the Oracle database:
apiVersion: apps/v1
kind: Deployment
metadata:
name: oracle-db
namespace: default
spec:
replicas: 1
selector:
matchLabels:
app.kubernetes.io/instance: dev
app.kubernetes.io/name: oracle-db
strategy:
rollingUpdate:
maxSurge: 1
maxUnavailable: 1
type: RollingUpdate
template:
metadata:
creationTimestamp: null
labels:
app.kubernetes.io/instance: dev
app.kubernetes.io/name: oracle-db
spec:
containers:
- env:
- name: DB_SID
value: devcdb
- name: DB_PDB
value: devpdb
- name: DB_DOMAIN
value: k8s
image: container-registry.oracle.com/database/enterprise:12.2.0.1-slim
imagePullPolicy: IfNotPresent
name: oracle-db
ports:
- containerPort: 1521
name: tns
protocol: TCP
resources:
limits:
cpu: "1"
memory: 2Gi
requests:
cpu: 200m
terminationMessagePath: /dev/termination-log
terminationMessagePolicy: File
dnsPolicy: ClusterFirst
restartPolicy: Always
schedulerName: default-scheduler
securityContext: {}
terminationGracePeriodSeconds: 30
Notice that you can pass in environment variables to set the SID, the name of the PDB, and
so on. The documentation describes the other variables that are available.
Follow the instructions in the documentation to set the sys password.
You should also create a service to make the database available within the Kubernetes cluster with a well known name. Here is an example:
apiVersion: v1
kind: Service
metadata:
name: oracle-db
namespace: default
spec:
ports:
- name: tns
port: 1521
protocol: TCP
targetPort: 1521
selector:
app.kubernetes.io/instance: dev
app.kubernetes.io/name: oracle-db
sessionAffinity: None
type: ClusterIP
In the example above, the database would be visible in the cluster using the address
oracle-db.default.svc.cluster.local:1521/devpdb.k8s.
When you run the database in the Kubernetes cluster, you will probably want to also
run RCU from a pod inside your network, though this
is not strictly necessary. You could create a NodePort to expose your database outside
the Kubernetes cluster and run RCU on another machine with access to the cluster.
If you wish to run the database outside Kubernetes, you need to create a way for containers running in pods in Kubernetes to see the database. This can be done by defining a Kubernetes Service with no selector and associating it with an endpoint definition, as shown in the example below:
kind: Service
apiVersion: v1
metadata:
name: database
spec:
type: ClusterIP
ports:
- port: 1521
targetPort: 1521
---
kind: Endpoints
apiVersion: v1
metadata:
name: database
subsets:
- addresses:
- ip: 129.123.1.4
ports:
- port: 1521
This creates a DNS name database in the current namespace, ordefault if no namespace is
specified, as in the example above. In this example, the fully qualified name would be
database.default.svc.cluster.local. The second part is the namespace.
If you looked up the ClusterIP for such a service, it would have an IP address on the overlay
network, that is the network inside the Kubernetes cluster. If you are using flannel,
for example, the address might be something like 10.0.1.25. Note that this is usually a
non-routed address.
From a container in a pod running in Kubernetes, you can make a connection to that address
and port 1521. Kubernetes will route the connection to the address provided in the
endpoint definition, in this example, 129.123.1.4:1521. This IP address (or name) is
resolved from the point of view of the Kubernetes Node’s IP stack, not the overlay network
inside the Kubernetes cluster. Note that this is a “real” routed IP address.
When you create your data sources, you would use the internal address, for example,
database:1521/some.service.
Because your database is externally accessible, you can run RCU in the normal way, from any machine on your network.
If you want to run RCU from a pod inside the Kubernetes cluster, you can use the container image that you built earlier as a “service” pod to run RCU. To do this, start up a pod using that image as follows:
$ kubectl run rcu --generator=run-pod/v1 --image container-registry.oracle.com/middleware/fmw_infrastructure:12.2.1.4 -- sleep infinity
This will create a Kubernetes Deployment called rcu containing a pod running a container
created from the container-registry.oracle.com/middleware/fmw_infrastructure:12.2.1.4 image which will just run
sleep infinity, which essentially creates a pod that we can “exec” into and use to run whatever
commands we need to run.
To get inside this container and run commands, use this command:
$ kubectl exec -ti rcu /bin/bash
When you are finished with this pod, you can remove it with this command:
$ kubectl delete pod rcu
You can use the same approach to get a temporary pod to run other utilities like WLST.
Inside this pod, you can use the following command to run RCU in command-line (no GUI) mode
to create your FMW schemas. You will need to provide the right prefix and connect string.
You will be prompted to enter the password for the sys user, and then the password to use
for the regular schema users:
$ /u01/oracle/oracle_common/bin/rcu \
-silent \
-createRepository \
-databaseType ORACLE \
-connectString oracle-db.default:1521/devpdb.k8s \
-dbUser sys \
-dbRole sysdba \
-useSamePasswordForAllSchemaUsers true \
-selectDependentsForComponents true \
-schemaPrefix FMW1 \
-component MDS \
-component IAU \
-component IAU_APPEND \
-component IAU_VIEWER \
-component OPSS \
-component WLS \
-component STB
You need to make sure that you maintain the association between the database schemas and the
matching domain just like you did in a non-Kubernetes environment. There is no specific
functionality provided to help with this. We recommend that you consider making the RCU
prefix (value of the schemaPrefix argument) the same as your domainUID to help maintain this association.
If you want to drop the schema, you can use a command like this:
$ /u01/oracle/oracle_common/bin/rcu \
-silent \
-dropRepository \
-databaseType ORACLE \
-connectString oracle-db.default:1521/devpdb.k8s \
-dbUser sys \
-dbRole sysdba \
-selectDependentsForComponents true \
-schemaPrefix FMW1 \
-component MDS \
-component IAU \
-component IAU_APPEND \
-component IAU_VIEWER \
-component OPSS \
-component WLS \
-component STB
Again, you will need to set the right prefix and connection string, and you will be prompted
to enter the sys user password.
You also need to create a Kubernetes Secret containing the credentials for the database schemas. When you create your domain using the sample provided below, it will obtain the RCU credentials from this secret.
We provide a sample
that demonstrates how to create the secret. The schema owner user name required will be the
schemaPrefix value followed by an underscore and a component name, such as FMW1_STB. The schema owner
password will be the password you provided for regular schema users during RCU creation.
Now that you have your images and you have created your RCU schemas, you are ready to create your domain. We provide a sample that demonstrates how to create an FMW Infrastructure domain.
There are two kinds of patches that can be applied to the FMW Infrastructure binaries:
You can find out whether or not a patch is eligible for Zero Downtime Patching by consulting the README file that accompanies the patch.
If you wish to apply a ZDP compliant patch which can be applied with a rolling restart, after you have patched the FMW Infrastructure image as shown in this sample, you can edit the domain custom resource with the name of the new image and the operator will initiate a rolling restart of the domain.
If you wish to apply a non-ZDP compliant patch to the FMW Infrastructure binary image, you must shut down the entire domain before applying the patch. Please see the documentation on domain lifecycle operations for more information.
An example of a non-ZDP compliant patch is one that includes a schema change that can not be applied dynamically.
If you are running a domain which contains Coherence, please refer to Coherence requirements for more information.