# Help setting up a Production sharded MongoDB Cluster with 3 replicas per shard

**URL:** <https://forums.percona.com/t/help-setting-up-a-production-sharded-mongodb-cluster-with-3-replicas-per-shard/36892>\
**Category:** Percona Server for MongoDB\
**Created:** [February 24, 2025, 12:28pm UTC](https://forums.percona.com/t/help-setting-up-a-production-sharded-mongodb-cluster-with-3-replicas-per-shard/36892 "2025-02-24T12:28:29Z")\
**Posts on this page:** 4\
**Page:** 1

<div class="post-metadata">

**Author:** ![Kay\_Khan](https://sea1.discourse-cdn.com/flex019/user_avatar/forums.percona.com/kay_khan/32/17023_2.png) [@Kay\_Khan](https://forums.percona.com/u/Kay_Khan)\
**Post date:** [February 24, 2025, 12:28pm UTC](https://forums.percona.com/t/help-setting-up-a-production-sharded-mongodb-cluster-with-3-replicas-per-shard/36892/1 "2025-02-24T12:28:29Z")

</div>

Hi we are looking for some help setting up a production ready cluster sharded with 3 replicas.

However we get the following error given the config below. Looking at this documentation -[MongoDB sharding - Percona Operator for MongoDB](https://docs.percona.com/percona-operator-for-mongodb/sharding.html) i can’t figure out what i am mising from out config.

Additionally looking at the error below and the helm values [percona-helm-charts/charts/psmdb-db at main · percona/percona-helm-charts · GitHub](https://github.com/percona/percona-helm-charts/tree/main/charts/psmdb-db) - replsets.cfg.members does not seem to be a valid property we can adjust

Bit confused what we are missing.

> 2025-02-24T12:15:17.143Z ERROR failed to update cluster status {“controller”: “psmdb-controller”, “controllerGroup”: “[psmdb.percona.com](http://psmdb.percona.com)”, “controllerKind”: “PerconaServerMongoDB”, “PerconaServerMongoDB”: {“name”:“psmdb-default-sharded”,“namespace”:“mongodb-sharded”}, “namespace”: “mongodb-sharded”, “name”: “psmdb-default-sharded”, “reconcileID”: “19eda940-1419-46ad-859a-270a2e21109b”, “replset”: “rs0”, “error”: “write status: [PerconaServerMongoDB.psmdb.percona.com](http://PerconaServerMongoDB.psmdb.percona.com) "psmdb-default-sharded" is invalid: status.replsets.cfg.members: Invalid value: "object": replsets.cfg.members in body must be of type array: "object"”, “errorVerbose”: “[PerconaServerMongoDB.psmdb.percona.com](http://PerconaServerMongoDB.psmdb.percona.com) "psmdb-default-sharded" is invalid: status.replsets.cfg.members: Invalid value: "object": replsets.cfg.members in body must be of type array: "object"\nwrite status\ngithub.com/percona/percona-server-mongodb-operator/pkg/controller/perconaservermongodb.(\*ReconcilePerconaServerMongoDB).writeStatus\n\t/go/src/github.com/percona/percona-server-mongodb-operator/pkg/controller/perconaservermongodb/status.go:269\ngithub.com/percona/percona-server-mongodb-operator/pkg/controller/perconaservermongodb.(\*ReconcilePerconaServerMongoDB).updateStatus\n\t/go/src/github.com/percona/percona-server-mongodb-operator/pkg/controller/perconaservermongodb/status.go:238\ngithub.com/percona/percona-server-mongodb-operator/pkg/controller/perconaservermongodb.(\*ReconcilePerconaServerMongoDB).Reconcile.func1\n\t/go/src/github.com/percona/percona-server-mongodb-operator/pkg/controller/perconaservermongodb/psmdb\_controller.go:266\ngithub.com/percona/percona-server-mongodb-operator/pkg/controller/perconaservermongodb.(\*ReconcilePerconaServerMongoDB).Reconcile\n\t/go/src/github.com/percona/percona-server-mongodb-operator/pkg/controller/perconaservermongodb/psmdb\_controller.go:470\nsigs.k8s.io/controller-runtime/pkg/internal/controller.(\*Controller[…]).Reconcile\n\t/go/pkg/mod/sigs.k8s.io/controller-runtime@v0.19.3/pkg/internal/controller/controller.go:116\nsigs.k8s.io/controller-runtime/pkg/internal/controller.(\*Controller[…]).reconcileHandler\n\t/go/pkg/mod/sigs.k8s.io/controller-runtime@v0.19.3/pkg/internal/controller/controller.go:303\nsigs.k8s.io/controller-runtime/pkg/internal/controller.(\*Controller[…]).processNextWorkItem\n\t/go/pkg/mod/sigs.k8s.io/controller-runtime@v0.19.3/pkg/internal/controller/controller.go:263\nsigs.k8s.io/controller-runtime/pkg/internal/controller.(\*Controller[…]).Start.func2.2\n\t/go/pkg/mod/sigs.k8s.io/controller-runtime@v0.19.3/pkg/internal/controller/controller.go:224\nruntime.goexit\n\t/usr/local/go/src/runtime/asm\_amd64.s:1700”}

```auto
finalizers:
  - delete-psmdb-pods-in-order
nameOverride: ${db_name}
unsafeFlags:
  replsetSize: true
upgradeOptions:
  apply: disabled
image:
  repository: percona/percona-server-mongodb
  tag: 8.0.4-1-multi
enableVolumeExpansion: true
replsets:
  rs0:
    name: rs0
    size: 3
    tolerations:
      - key: "karpenter/mongodb"
        operator: "Exists"
        effect: "NoSchedule"
    nodeSelector:
      karpenter-node-pool: mongodb
    resources: 
      limits:
        cpu: 2
        memory: 4Gi
      requests:
        cpu: 300m
        memory: 500M
    expose:
      enabled: false
    volumeSpec:
      pvc:
        storageClassName: mongodb
        resources:
          requests:
            storage: 100Gi
  rs1:
    name: rs1
    size: 3
    tolerations:
      - key: "karpenter/mongodb"
        operator: "Exists"
        effect: "NoSchedule"
    nodeSelector:
      karpenter-node-pool: mongodb
    resources: 
      limits:
        cpu: 2
        memory: 4Gi
      requests:
        cpu: 300m
        memory: 500M
    expose:
      enabled: false
    volumeSpec:
      pvc:
        storageClassName: mongodb
        resources:
          requests:
            storage: 100Gi
  rs2:
    name: rs2
    size: 3
    tolerations:
      - key: "karpenter/mongodb"
        operator: "Exists"
        effect: "NoSchedule"
    nodeSelector:
      karpenter-node-pool: mongodb
    resources: 
      limits:
        cpu: 2
        memory: 4Gi
      requests:
        cpu: 300m
        memory: 500M
    expose:
      enabled: false
    volumeSpec:
      pvc:
        storageClassName: mongodb
        resources:
          requests:
            storage: 100Gi
backup:
  enabled: false
sharding:
  enabled: true

```

---

<div class="post-metadata">

**Author:** ![radoslaw.szulgo](https://sea1.discourse-cdn.com/flex019/user_avatar/forums.percona.com/radoslaw.szulgo/32/17315_2.png) [@radoslaw.szulgo](https://forums.percona.com/u/radoslaw.szulgo)\
**Post date:** [February 24, 2025, 12:32pm UTC](https://forums.percona.com/t/help-setting-up-a-production-sharded-mongodb-cluster-with-3-replicas-per-shard/36892/2 "2025-02-24T12:32:40Z")

</div>

Looks similar to this one:  
[https://forums.percona.com/t/error-upgrading-to-1-19-failed-to-update-cluster-status/36422/4](https://forums.percona.com/t/error-upgrading-to-1-19-failed-to-update-cluster-status/36422/4)

---

<div class="post-metadata">

**Author:** ![Kay\_Khan](https://sea1.discourse-cdn.com/flex019/user_avatar/forums.percona.com/kay_khan/32/17023_2.png) [@Kay\_Khan](https://forums.percona.com/u/Kay_Khan)\
**Post date:** [February 24, 2025, 12:36pm UTC](https://forums.percona.com/t/help-setting-up-a-production-sharded-mongodb-cluster-with-3-replicas-per-shard/36892/3 "2025-02-24T12:36:30Z")

</div>

Interesting, maybe you could shed some light on what the best practise is here.

We have a test eks cluser - where i originally installed just one operator + db.

1. **namespace: mongodb - percona operator + db: version 1.18**

I am now looking at testing out a sharded clsuter and have installed it onto

1. **namespace: mongodb-sharded - percona operator + db: version 1.19.1**

So could it be that despite them being in separate namespaces, the crds originally installed in 1. is being used for 2. What is the best way to keep them separate?

```auto
psmdb-operator mongodb-sharded 1 2025-02-24 11:42:54.263168754 +0000 UTC deployed psmdb-operator-1.19.1 1.19.1
psmdb-operator mongodb 1 2025-01-02 12:07:12.658132847 +0000 UTC deployed psmdb-operator-1.18.0 1.18.0

```

aside: is there a seperate helm chart for crds?

---

<div class="post-metadata">

**Author:** ![radoslaw.szulgo](https://sea1.discourse-cdn.com/flex019/user_avatar/forums.percona.com/radoslaw.szulgo/32/17315_2.png) [@radoslaw.szulgo](https://forums.percona.com/u/radoslaw.szulgo)\
**Post date:** [September 10, 2025, 7:11pm UTC](https://forums.percona.com/t/help-setting-up-a-production-sharded-mongodb-cluster-with-3-replicas-per-shard/36892/5 "2025-09-10T19:11:19Z")

</div>

Yes, your suspicion is absolutely correct. Despite being in separate namespaces, the CRDs from the older installation are affecting the newer operator installation.

The core issue is that **CRDs are cluster-scoped resources, not namespaced**.

The standard and highly recommended best practice is to run only **one version of a specific operator per cluster**. This ensures there is a single, unambiguous CRD version that all instances of the custom resource adhere to. To keep your database environments separate, continue using namespaces, but manage them all with a single, centrally managed operator.

####