Docs Menu
Docs Home
/
MongoDB Enterprise Kubernetes Operator
/

Configure an Ops Manager Resource to use Local Mode

On this page

Important

Configuring Ops Manager to use Local Mode in Kubernetes is not recommended. Consider configuring Ops Manager to use Remote Mode instead.

In a default configuration, the MongoDB Agents and Backup Daemons access MongoDB installation archives over the Internet from MongoDB, Inc.

You can configure Ops Manager to run in Local Mode with the Kubernetes Operator if the nodes in your Kubernetes cluster don't have access to the Internet. The Backup Daemons and managed MongoDB resources download installation archives only from a Persistent Volume that you create for the Ops Manager StatefulSet.

This procedure covers uploading installation archives to Ops Manager.

If Ops Manager has no internet access, see Configure Deployment to Have Limited Internet Access.

For compatbility, see MongoDB Enterprise Kubernetes Operator Compatibility. To view all available versions for each image, see Container Images.

  • Deploy an Ops Manager Resource. The following procedure shows you how to update your Ops Manager Kubernetes object to enable Local Mode.

  • To avoid downtime when you enable Local Mode, ensure that you set spec.replicas to a value greater than 1 in your Ops Manager resource definition.

    If you updated your Ops Manager resource definition to make Ops Manager highly available, apply your changes before you begin this tutorial:

    kubectl apply -f <opsmgr-resource>.yaml -n <metadata.namespace>
1

If you have not already, run the following command to execute all kubectl commands in the namespace you created.

Note

If you are deploying an Ops Manager resource in a multi-Kubernetes cluster MongoDB deployment:

  • Set the context to the name of the central cluster, such as: kubectl config set context "$MDB_CENTRAL_CLUSTER_FULL_NAME".

  • Set the --namespace to the same scope that you used for your multi-Kubernetes cluster MongoDB deployment, such as: kubectl config --namespace "mongodb".

kubectl config set-context $(kubectl config current-context) --namespace=<metadata.namespace>
2

In this tutorial, you update the StatefulSet that manages the Ops Manager Pods in your Kubernetes cluster.

You must first delete the Ops Manager StatefulSet so that Kubernetes can apply the updates that Local Mode requires.

  1. Find the name of your Ops Manager StatefulSet:

    kubectl get statefulsets

    The entry in the response that matches the metadata.name of your

    Your Ops Manager StatefulSet is the entry in the response that matches the metadata.name in your Ops Manager resource definition.

    kubectl get statefulsets -n mongodb
    NAME READY AGE
    ops-manager-localmode 2/2 2m31s
    ops-manager-localmode-db 3/3 4m46s
  2. Delete the Ops Manager StatefulSet:

    Warning

    Ensure that you include the --cascade=false flag when you delete your Ops Manager StatefulSet. If you don't include this flag, Kubernetes also deletes your Ops Manager Pods.

    kubectl delete statefulset --cascade=false <ops-manager-statefulset>
3

Copy Lines 9-31 of this example to:

  • Use the Ops Manager configuration setting automation.versions.source: local in spec.configuration to enable Local Mode.

  • Define a Persistent Volume for the Ops Manager StatefulSet to store the MongoDB installation archive. MongoDB Agents running in MongoDB database resource containers that you create with the Kubernetes Operator download the installation archives from Ops Manager instead of from the Internet.

1apiVersion: mongodb.com/v1
2kind: MongoDBOpsManager
3metadata:
4 name: ops-manager-localmode
5spec:
6 replicas: 2
7 version: "6.0.0"
8 adminCredentials: ops-manager-admin-secret
9 configuration:
10 # this enables local mode in Ops Manager
11 automation.versions.source: local
12 statefulSet:
13 spec:
14 # the Persistent Volume Claim will be created for each Ops Manager Pod
15 volumeClaimTemplates:
16 - metadata:
17 name: mongodb-versions
18 spec:
19 accessModes: [ "ReadWriteOnce" ]
20 resources:
21 requests:
22 storage: "20Gi"
23 template:
24 spec:
25 containers:
26 - name: mongodb-ops-manager
27 volumeMounts:
28 - name: mongodb-versions
29 # this is the directory in each Pod where all MongoDB
30 # archives must be put
31 mountPath: /mongodb-ops-manager/mongodb-releases
32 backup:
33 enabled: false
34 applicationDatabase:
35 members: 3
4

Open your preferred text editor and paste the object specification into the appropriate location in your resource file.

5
6
  1. Invoke the following kubectl command on the filename of the Ops Manager resource definition:

    kubectl apply -f <opsmgr-resource>.yaml
  2. Kubernetes creates a new Ops Manager StatefulSet when you apply the changes to your Ops Manager resource definition. Before proceeding to the next step, run the following command to ensure that the Ops Manager StatefulSet exists:

    kubectl get statefulsets

    The new Ops Manager StatefulSet should show 0 members ready:

    kubectl get statefulsets -n mongodb
    NAME READY AGE ops-manager-localmode 0/2 2m31s
    ops-manager-localmode-db 3/3 4m46s
7
  1. List the Ops Manager Pods in your Kubernetes cluster:

    kubectl get pods
  2. Delete one Ops Manager Pod:

    kubectl delete pod <om-pod-0>
  3. Kubernetes recreates the Ops Manager Pod you deleted. Continue to get the status of the new Pod until it is ready:

    kubectl get pods

    When the new Pod is initializing, the output is similar to the following example:

    NAME READY STATUS RESTARTS AGE
    mongodb-enterprise-operator-5648d4c86-k5brh 1/1 Running 0 5m24s
    ops-manager-localmode-0 0/1 Running 0 0m55s
    ops-manager-localmode-1 1/1 Running 0 5m45s
    ops-manager-localmode-db-0 1/1 Running 0 5m19s
    ops-manager-localmode-db-1 1/1 Running 0 4m54s
    ops-manager-localmode-db-2 1/1 Running 0 4m12s

    When the new Pod is ready, the output is similar to the following example:

    NAME READY STATUS RESTARTS AGE
    mongodb-enterprise-operator-5648d4c86-k5brh 1/1 Running 0 5m24s
    ops-manager-localmode-0 1/1 Running 0 3m55s
    ops-manager-localmode-1 1/1 Running 0 5m45s
    ops-manager-localmode-db-0 1/1 Running 0 5m19s
    ops-manager-localmode-db-1 1/1 Running 0 4m54s
    ops-manager-localmode-db-2 1/1 Running 0 4m12s
  4. Repeat Steps b and c until you've deleted all of your Ops Manager Pods and confirmed that all of the new Pods are ready.

8

To check the status of your Ops Manager resource, invoke the following command:

kubectl get om -o yaml -w

See Troubleshoot the Kubernetes Operator for information about the resource deployment statuses.

After the Ops Manager resource completes the Pending phase, the command returns output similar to the following:

1status:
2 applicationDatabase:
3 lastTransition: "2020-05-15T16:20:22Z"
4 members: 3
5 phase: Running
6 type: ReplicaSet
7 version: "4.4.5-ubi8"
8 backup:
9 phase: ""
10 opsManager:
11 lastTransition: "2020-05-15T16:20:26Z"
12 phase: Running
13 replicas: 1
14 url: http://ops-manager-localmode-svc.mongodb.svc.cluster.local:8080
15 version: "6.0.0"

Copy the value of the status.opsManager.url field, which states the resource's connection URL. You use this value when you create a ConfigMap later in the procedure.

9

The installers that you download depend on the environment to which you deployed the operator:

Note

The following example includes a link that allows you to download the specified version of MongoDB Community Edition. To download any other version of MongoDB Community Edition, visit the MongoDB Community Edition Download Center. To download MongoDB Enterprise Edition, visit the MongoDB Enterprise Download Center.

Download the RHEL installation tarball for the MongoDB Server version you want the Kubernetes Operator to deploy. For example, to download the 6.0.1 release:

curl -OL https://fastdl.mongodb.org/linux/mongodb-linux-x86_64-rhel80-6.0.1.tgz
10

Copy the MongoDB archive for each MongoDB version you intend to deploy to the Ops Manager Persistent Volume.

The commands that you use depend on the environment to which you deployed the Kubernetes Operator:

Note

If you deployed more than one Ops Manager replica, copy only the MongoDB installation tarball packages to Replica 1 and beyond.

11
  1. If you have not done so already, complete the following prerequisites:

  2. Deploy a MongoDB database resource in the same namespace to which you deployed Ops Manager. Ensure that you:

    1. Match the spec.opsManager.configMapRef.name of the resource to the metadata.name of your ConfigMap.

    2. Match the spec.credentials of the resource to the name of the secret you created that contains an Ops Manager programmatic API key pair.

MongoDB Agents run in MongoDB database resource containers that you create with the Kubernetes Operator. Download the installation archives from Ops Manager instead of downloading them from the Internet.

← Configure an