Navigation
This version of the documentation is archived and no longer supported.

Deploy a Sharded Cluster

Note

At any place on this page that says Ops Manager, you can substitute Cloud Manager.

Important

  • You can use the Kubernetes Operator to deploy MongoDB resources with Ops Manager version 4.0.11 or later and Cloud Manager.
  • You can’t use the Kubernetes Operator to deploy MongoDB resources to Atlas.

Sharded clusters provide horizontal scaling for large data sets and enable high throughput operations by distributing the data set across a group of servers.

To learn more about sharding, see Sharding Introduction in the MongoDB manual.

Use this procedure to deploy a new sharded cluster that Ops Manager manages. Later, you can use Ops Manager to add shards and perform other maintenance operations on the cluster.

Prerequisites

To deploy a sharded cluster using an object, you need to complete the following procedures:

Considerations

Starting in MongoDB Enterprise Kubernetes Operator version 1.3.0, you can only have one MongoDB resource per project. To learn how to migrate your project to a single-cluster configuration, see Migrate to One Resource per Project (Required for Version 1.3.0).

Do Not Deploy Monitoring Agents inside and outside Kubernetes

Do not mix MongoDB deployments outside Kubernetes with ones inside Kubernetes in the same Project.

Due to Kubernetes network translation, a Monitoring Agent outside Kubernetes cannot monitor MongoDB instances inside Kubernetes. For this reason, k8s and non-k8s deployments in the same Project are not supported. Use separate projects.

Procedure

The procedure for deploying a sharded cluster depends on whether you require the deployment to run with TLS enabled for intra-cluster communication and clients connecting to the database:

1

Configure kubectl to default to your namespace.

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

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

Copy the highlighted section of this sharded cluster resource.

Change the highlighted settings of this YAML file to match your desired sharded cluster configuration.

This is a YAML file that you can modify to meet your desired configuration. Change the highlighted settings to match your desired sharded cluster configuration.

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
---
apiVersion: mongodb.com/v1
kind: MongoDB
metadata:
  name: <my-sharded-cluster>
spec:
  shardCount: 2
  mongodsPerShardCount: 3
  mongosCount: 2
  configServerCount: 3
  version: "4.2.2-ent"
  opsManager:
    configMapRef:
      name: <configMap.metadata.name>
            # Must match metadata.name in ConfigMap file
  credentials: <mycredentials>
  type: ShardedCluster
  persistent: true
...
3

Paste the copied example to create a new sharded cluster resource.

Open your preferred text editor and paste the object specification into a new text file.

4

Configure the settings highlighted in the preceding step as follows.

Key Type Description Example
metadata.name string

Label for this Kubernetes sharded cluster object.

Resource names must be 44 characters or less.

See also

myproject
spec.shardCount integer Number of shards to deploy. 2
spec.mongodsPerShardCount integer Number of shard members per shard. 3
spec.mongosCount integer Number of shard routers to deploy. 2
spec.configServerCount integer Number of members of the config server replica set. 3
spec.version string

Version of MongoDB that this sharded cluster should run.

The format should be X.Y.Z for the Community edition and X.Y.Z-ent for the Enterprise edition.

Important

Ensure that you choose a compatible MongoDB Server version.

Compatible versions differ depending on the base image that the MongoDB database resource uses.

To learn more about MongoDB versioning, see MongoDB Versioning in the MongoDB Manual.

4.2.11-ent
spec.opsManager.configMapRef.name string

Name of the ConfigMap with the Ops Manager connection configuration. The spec.cloudManager.configMapRef.name setting is an alias for this setting and can be used in its place.

Value must match namespace and name of ConfigMap

This value must match the namespace in which you created the Ops Manager project ConfigMap.

Operator manages changes to the ConfigMap

The Kubernetes Operator tracks any changes to the ConfigMap and reconciles the state of the MongoDB Kubernetes resource.

<myproject>
spec.credentials string

Name of the Kubernetes secret you created as Ops Manager API authentication credentials for the Kubernetes Operator to communicate with Ops Manager.

Value must use namespace and name of Secret

This value must match the namespace in which you created the secret and the name value you provided for your Ops Manager Kubernetes Secret.

Operator manages changes to the Secret

The Kubernetes Operator tracks any changes to the Secret and reconciles the state of the MongoDB Kubernetes resource.

<mycredentials>
spec.type string Type of MongoDB Kubernetes resource to create. ShardedCluster
spec.persistent string

Optional.

Flag indicating if this MongoDB Kubernetes resource should use Persistent Volumes for storage. Persistent volumes are not deleted when the MongoDB Kubernetes resource is stopped or restarted.

If this value is true, then the following values are set to their default value of 16Gi:

To change your Persistent Volume Claims configuration, configure the following collections to meet your deployment requirements:

Warning

Your containers must have permissions to write to your Persistent Volume. The Kubernetes Operator sets fsGroup = 2000 in securityContext This makes Kubernetes try to fix write permissions for the Persistent Volume. If redeploying the deployment item does not fix issues with your Persistent Volumes, contact MongoDB Support.

Note

If you do not use Persistent Volumes, the Disk Usage and Disk IOPS charts cannot be displayed in either the Processes tab on the Deployment page or in the Metrics page when reviewing the data for this deployment.

true
5

Add any additional accepted settings for a sharded cluster deployment.

You can also add any of the following optional settings to the object specification file for a sharded cluster deployment:

Warning

You must set spec.clusterDomain if your Kubernetes cluster has a default domain other than the default cluster.local. If you neither use the default nor set the spec.clusterDomain option, the Kubernetes Operator might not function as expected.

For config server

For shard routers

For shard members

6

Save this file with a .yaml file extension.

7

Start your sharded cluster deployment.

Invoke the following Kubernetes command to create your sharded cluster:

kubectl apply -f <sharded-cluster-conf>.yaml

Check the log after running this command. If the creation was successful, you should see a message similar to the following:

2018-06-26T10:30:30.346Z INFO operator/shardedclusterkube.go:52 Created! {"sharded cluster": "my-sharded-cluster"}
8

Track the status of your sharded cluster deployment.

To check the status of your MongoDB Kubernetes resource, invoke the following command:

kubectl get mdb <resource-name> -o yaml -w

The -w flag means “watch”. With the “watch” flag set, the output refreshes immediately when the configuration changes until the status phase achieves the Running state.

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