Milvus Quickstart

This guide provides a comprehensive walkabout for deploying and managing Milvus ReplicaSet Clusters using the KubeBlocks Milvus Add-on, covering:

System prerequisites and add-on installation
Cluster creation and configuration
Operational management including start/stop procedures
Connection methods and cluster monitoring

Prerequisites

System Requirements

Before proceeding, verify your environment meets these requirements:

A functional Kubernetes cluster (v1.21+ recommended)
kubectl v1.21+ installed and configured with cluster access
Helm installed (installation guide)
KubeBlocks installed (installation guide)

Verify Milvus Add-on

The Milvus Add-on is included with KubeBlocks by default. Check its status:

helm list -n kb-system | grep milvus

Example Output:

NAME                  NAMESPACE   REVISION    UPDATED                     STATUS      CHART
kb-addon-milvus       kb-system   1           2025-05-21                  deployed    milvus-1.0.0

If the add-on isn't enabled, choose an installation method:


# Add Helm repo
helm repo add kubeblocks-addons https://apecloud.github.io/helm-charts
# For users in Mainland China, if GitHub is inaccessible or slow, use this alternative repo:
#helm repo add kubeblocks-addons https://jihulab.com/api/v4/projects/150246/packages/helm/stable

# Update helm repo
helm repo update
# Search available Add-on versions
helm search repo kubeblocks/milvus --versions
# Install your desired version (replace <VERSION> with your chosen version)
helm upgrade -i kb-addon-milvus kubeblocks-addons/milvus --version <VERSION> -n kb-system


# Add an index (kubeblocks is added by default)
kbcli addon index add kubeblocks https://github.com/apecloud/block-index.git
# Update the index
kbcli addon index update kubeblocks
# Update all indexes
kbcli addon index update --all

To search and install an addon:


# Search Add-on
kbcli addon search milvus
# Install Add-on with your desired version (replace <VERSION> with your chosen version)
kbcli addon install milvus --version <VERSION>

Example Output:

ADDON   VERSION          INDEX
milvus   0.9.0           kubeblocks
milvus   0.9.1           kubeblocks
milvus   1.0.0           kubeblocks

To enable or disable an addon:


# Enable Add-on
kbcli addon enable milvus
# Disable Add-on
kbcli addon disable milvus

NOTE

Version Compatibility

Always verify that the Milvus Add-on version matches your KubeBlocks major version to avoid compatibility issues.

Verify Supported Milvus Versions

List available Milvus versions:

kubectl get cmpv milvus

Example Output

NAME     VERSIONS   STATUS      AGE
milvus   v2.3.2     Available   26d

Storage Configuration

Milvus requires persistent storage. Verify available options:

kubectl get storageclass

Recommended storage characteristics:

Minimum 20Gi capacity
ReadWriteOnce access mode
Supports volume expansion
Appropriate performance for workload

Deploy a Milvus Cluster

Deploy a basic Milvus Cluster with default settings:

kubectl apply -f https://raw.githubusercontent.com/apecloud/kubeblocks-addons/refs/heads/main/examples/milvus/cluster-standalone.yaml

This creates:

A Milvus Cluster with 3 replicas, one for milvus, one for etcd and one for minio.
Default resource allocations (0.5 CPU, 0.5Gi memory)
20Gi persistent storage


apiVersion: apps.kubeblocks.io/v1
kind: Cluster
metadata:
  name: milvus-standalone
  namespace: demo
spec:
  # Specifies the behavior when a Cluster is deleted.
  # Valid options are: [DoNotTerminate, Delete, WipeOut] (`Halt` is deprecated since KB 0.9)
  # - `DoNotTerminate`: Prevents deletion of the Cluster. This policy ensures that all resources remain intact.
  # - `Delete`: Extends the `Halt` policy by also removing PVCs, leading to a thorough cleanup while removing all persistent data.
  # - `WipeOut`: An aggressive policy that deletes all Cluster resources, including volume snapshots and backups in external storage. This results in complete data removal and should be used cautiously, primarily in non-production environments to avoid irreversible data loss.
  terminationPolicy: Delete
  # Specifies the name of the ClusterDefinition to use when creating a Cluster.
  # Note: DO NOT UPDATE THIS FIELD
  # The value must be `milvus` to create a Milvus Cluster
  clusterDef: milvus
  # Specifies the name of the ClusterTopology to be used when creating the
  # Cluster.
  # Valid options are: [standalone,cluster]
  topology: standalone
  # Specifies a list of ClusterComponentSpec objects used to define the
  # individual Components that make up a Cluster.
  # This field allows for detailed configuration of each Component within the Cluster
  componentSpecs:
    - name: etcd
      replicas: 1
      resources:
        limits:
          cpu: "0.5"
          memory: "0.5Gi"
        requests:
          cpu: "0.5"
          memory: "0.5Gi"
      volumeClaimTemplates:
        - name: data
          spec:
            storageClassName: ""
            accessModes:
              - ReadWriteOnce
            resources:
              requests:
                storage: 20Gi
    - name: minio
      replicas: 1
      resources:
        limits:
          cpu: "0.5"
          memory: "0.5Gi"
        requests:
          cpu: "0.5"
          memory: "0.5Gi"
      volumeClaimTemplates:
        - name: data
          spec:
            storageClassName: ""
            accessModes:
              - ReadWriteOnce
            resources:
              requests:
                storage: 20Gi
    - name: milvus
      replicas: 1
      resources:
        limits:
          cpu: "0.5"
          memory: "0.5Gi"
        requests:
          cpu: "0.5"
          memory: "0.5Gi"
      volumeClaimTemplates:
        - name: data
          spec:
            storageClassName: ""
            accessModes:
              - ReadWriteOnce
            resources:
              requests:
                storage: 20Gi

For more API fields and descriptions, refer to the API Reference.

Verify Cluster Status

When deploying a Milvus Cluster with 3 replicas:

Confirm successful deployment by checking:

Cluster phase is Running
All pods are operational

Check status using either method:

kubectl get cluster milvus-standalone -n demo -w
NAME                CLUSTER-DEFINITION   TERMINATION-POLICY   STATUS     AGE
milvus-standalone   milvus               Delete               Creating   27s
milvus-standalone   milvus               Delete               Running    64s

kubectl get pods -l app.kubernetes.io/instance=milvus-standalone -n demo
NAME                         READY   STATUS    RESTARTS   AGE
milvus-standalone-etcd-0     2/2     Running   0          25m
milvus-standalone-milvus-0   1/1     Running   0          24m
milvus-standalone-minio-0    1/1     Running   0          25m

With kbcli installed, you can view comprehensive cluster information:

kbcli cluster describe milvus-standalone -n demo

Name: milvus-standalone	 Created Time: May 19,2025 11:03 UTC+0800
NAMESPACE   CLUSTER-DEFINITION   TOPOLOGY     STATUS    TERMINATION-POLICY
demo        milvus               standalone   Running   Delete

Endpoints:
COMPONENT   INTERNAL   EXTERNAL

Topology:
COMPONENT   SERVICE-VERSION   INSTANCE                     ROLE     STATUS    AZ       NODE    CREATED-TIME
etcd        3.5.15            milvus-standalone-etcd-0     leader   Running   zone-x   x.y.z   May 19,2025 11:03 UTC+0800
milvus      v2.3.2            milvus-standalone-milvus-0   <none>   Running   zone-x   x.y.z   May 19,2025 11:04 UTC+0800
minio       8.0.17            milvus-standalone-minio-0    <none>   Running   zone-x   x.y.z   May 19,2025 11:03 UTC+0800

Resources Allocation:
COMPONENT   INSTANCE-TEMPLATE   CPU(REQUEST/LIMIT)   MEMORY(REQUEST/LIMIT)   STORAGE-SIZE   STORAGE-CLASS
etcd                            500m / 500m          512Mi / 512Mi           data:10Gi      <none>
minio                           500m / 500m          512Mi / 512Mi           data:10Gi      <none>
milvus                          500m / 500m          512Mi / 512Mi           data:10Gi      <none>

Images:
COMPONENT   COMPONENT-DEFINITION              IMAGE
etcd        etcd-3-1.0.0                      quay.io/coreos/etcd:v3.5.15
minio       milvus-minio-1.0.0                docker.io/minio/minio:RELEASE.2022-03-17T06-34-49Z
milvus      milvus-standalone-1.0.0           docker.io/milvusdb/milvus:v2.3.2

Data Protection:
BACKUP-REPO   AUTO-BACKUP   BACKUP-SCHEDULE   BACKUP-METHOD   BACKUP-RETENTION   RECOVERABLE-TIME

Show cluster events: kbcli cluster list-events -n demo milvus-standalone

Access Milvus

To access the Milvus service, you can expose the service by creating a service:

kubectl port-forward pod/milvus-standalone-milvus-0 -n demo 19530:19530

And then you can access the Milvus service via localhost:19530.

Stop the Milvus Cluster

Stopping a cluster temporarily suspends operations while preserving all data and configuration:

Key Effects:

Compute resources (Pods) are released
Persistent storage (PVCs) remains intact
Service definitions are maintained
Cluster configuration is preserved
Operational costs are reduced


apiVersion: operations.kubeblocks.io/v1alpha1
kind: OpsRequest
metadata:
  name: milvus-stop
  namespace: demo
spec:
  clusterName: milvus-standalone
  type: Stop

Alternatively, stop by setting spec.componentSpecs.stop to true:


kubectl patch cluster milvus-standalone -n demo --type='json' -p='[
{
  "op": "add",
  "path": "/spec/componentSpecs/0/stop",
  "value": true
},
{
  "op": "add",
  "path": "/spec/componentSpecs/1/stop",
  "value": true
},
{
  "op": "add",
  "path": "/spec/componentSpecs/2/stop",
  "value": true
}
]'

Start the Milvus Cluster

Restarting a stopped cluster resumes operations with all data and configuration intact.

Key Effects:

Compute resources (Pods) are recreated
Services become available again
Cluster returns to previous state


apiVersion: operations.kubeblocks.io/v1alpha1
kind: OpsRequest
metadata:
  name: milvus-start
  namespace: demo
spec:
  clusterName: milvus-standalone
  type: Start

Restart by setting spec.componentSpecs.stop to false:


kubectl patch cluster milvus-standalone -n demo --type='json' -p='[
{
  "op": "remove",
  "path": "/spec/componentSpecs/0/stop"
},
{
  "op": "remove",
  "path": "/spec/componentSpecs/1/stop"
},
{
  "op": "remove",
  "path": "/spec/componentSpecs/2/stop"
}
]'

Delete Milvus Cluster

Choose carefully based on your data retention needs:

Policy	Resources Removed	Data Removed	Recommended For
DoNotTerminate	None	None	Critical production clusters
Delete	All resources	PVCs deleted	Non-critical environments
WipeOut	All resources	Everything*	Test environments only

*Includes snapshots and backups in external storage

Pre-Deletion Checklist:

Verify no applications are using the cluster
Ensure required backups exist
Confirm proper terminationPolicy is set
Check for dependent resources

For test environments, use this complete cleanup:


kubectl patch cluster milvus-standalone -p '{"spec":{"terminationPolicy":"WipeOut"}}' --type="merge" -n demo
kubectl delete cluster milvus-standalone -n demo

Milvus Quickstart

This guide provides a comprehensive walkabout for deploying and managing Milvus ReplicaSet Clusters using the KubeBlocks Milvus Add-on, covering:

System prerequisites and add-on installation
Cluster creation and configuration
Operational management including start/stop procedures
Connection methods and cluster monitoring

Prerequisites

System Requirements

Before proceeding, verify your environment meets these requirements:

A functional Kubernetes cluster (v1.21+ recommended)
kubectl v1.21+ installed and configured with cluster access
Helm installed (installation guide)
KubeBlocks installed (installation guide)

Verify Milvus Add-on

The Milvus Add-on is included with KubeBlocks by default. Check its status:

helm list -n kb-system | grep milvus

Example Output:

NAME                  NAMESPACE   REVISION    UPDATED                     STATUS      CHART
kb-addon-milvus       kb-system   1           2025-05-21                  deployed    milvus-1.0.0

If the add-on isn't enabled, choose an installation method:


# Add Helm repo
helm repo add kubeblocks-addons https://apecloud.github.io/helm-charts
# For users in Mainland China, if GitHub is inaccessible or slow, use this alternative repo:
#helm repo add kubeblocks-addons https://jihulab.com/api/v4/projects/150246/packages/helm/stable

# Update helm repo
helm repo update
# Search available Add-on versions
helm search repo kubeblocks/milvus --versions
# Install your desired version (replace <VERSION> with your chosen version)
helm upgrade -i kb-addon-milvus kubeblocks-addons/milvus --version <VERSION> -n kb-system


# Add an index (kubeblocks is added by default)
kbcli addon index add kubeblocks https://github.com/apecloud/block-index.git
# Update the index
kbcli addon index update kubeblocks
# Update all indexes
kbcli addon index update --all

To search and install an addon:


# Search Add-on
kbcli addon search milvus
# Install Add-on with your desired version (replace <VERSION> with your chosen version)
kbcli addon install milvus --version <VERSION>

Example Output:

ADDON   VERSION          INDEX
milvus   0.9.0           kubeblocks
milvus   0.9.1           kubeblocks
milvus   1.0.0           kubeblocks

To enable or disable an addon:


# Enable Add-on
kbcli addon enable milvus
# Disable Add-on
kbcli addon disable milvus

NOTE

Version Compatibility

Always verify that the Milvus Add-on version matches your KubeBlocks major version to avoid compatibility issues.

Verify Supported Milvus Versions

List available Milvus versions:

kubectl get cmpv milvus

Example Output

NAME     VERSIONS   STATUS      AGE
milvus   v2.3.2     Available   26d

Storage Configuration

Milvus requires persistent storage. Verify available options:

kubectl get storageclass

Recommended storage characteristics:

Minimum 20Gi capacity
ReadWriteOnce access mode
Supports volume expansion
Appropriate performance for workload

Deploy a Milvus Cluster

Deploy a basic Milvus Cluster with default settings:

kubectl apply -f https://raw.githubusercontent.com/apecloud/kubeblocks-addons/refs/heads/main/examples/milvus/cluster-standalone.yaml

This creates:

A Milvus Cluster with 3 replicas, one for milvus, one for etcd and one for minio.
Default resource allocations (0.5 CPU, 0.5Gi memory)
20Gi persistent storage


apiVersion: apps.kubeblocks.io/v1
kind: Cluster
metadata:
  name: milvus-standalone
  namespace: demo
spec:
  # Specifies the behavior when a Cluster is deleted.
  # Valid options are: [DoNotTerminate, Delete, WipeOut] (`Halt` is deprecated since KB 0.9)
  # - `DoNotTerminate`: Prevents deletion of the Cluster. This policy ensures that all resources remain intact.
  # - `Delete`: Extends the `Halt` policy by also removing PVCs, leading to a thorough cleanup while removing all persistent data.
  # - `WipeOut`: An aggressive policy that deletes all Cluster resources, including volume snapshots and backups in external storage. This results in complete data removal and should be used cautiously, primarily in non-production environments to avoid irreversible data loss.
  terminationPolicy: Delete
  # Specifies the name of the ClusterDefinition to use when creating a Cluster.
  # Note: DO NOT UPDATE THIS FIELD
  # The value must be `milvus` to create a Milvus Cluster
  clusterDef: milvus
  # Specifies the name of the ClusterTopology to be used when creating the
  # Cluster.
  # Valid options are: [standalone,cluster]
  topology: standalone
  # Specifies a list of ClusterComponentSpec objects used to define the
  # individual Components that make up a Cluster.
  # This field allows for detailed configuration of each Component within the Cluster
  componentSpecs:
    - name: etcd
      replicas: 1
      resources:
        limits:
          cpu: "0.5"
          memory: "0.5Gi"
        requests:
          cpu: "0.5"
          memory: "0.5Gi"
      volumeClaimTemplates:
        - name: data
          spec:
            storageClassName: ""
            accessModes:
              - ReadWriteOnce
            resources:
              requests:
                storage: 20Gi
    - name: minio
      replicas: 1
      resources:
        limits:
          cpu: "0.5"
          memory: "0.5Gi"
        requests:
          cpu: "0.5"
          memory: "0.5Gi"
      volumeClaimTemplates:
        - name: data
          spec:
            storageClassName: ""
            accessModes:
              - ReadWriteOnce
            resources:
              requests:
                storage: 20Gi
    - name: milvus
      replicas: 1
      resources:
        limits:
          cpu: "0.5"
          memory: "0.5Gi"
        requests:
          cpu: "0.5"
          memory: "0.5Gi"
      volumeClaimTemplates:
        - name: data
          spec:
            storageClassName: ""
            accessModes:
              - ReadWriteOnce
            resources:
              requests:
                storage: 20Gi

For more API fields and descriptions, refer to the API Reference.

Verify Cluster Status

When deploying a Milvus Cluster with 3 replicas:

Confirm successful deployment by checking:

Cluster phase is Running
All pods are operational

Check status using either method:

kubectl get cluster milvus-standalone -n demo -w
NAME                CLUSTER-DEFINITION   TERMINATION-POLICY   STATUS     AGE
milvus-standalone   milvus               Delete               Creating   27s
milvus-standalone   milvus               Delete               Running    64s

kubectl get pods -l app.kubernetes.io/instance=milvus-standalone -n demo
NAME                         READY   STATUS    RESTARTS   AGE
milvus-standalone-etcd-0     2/2     Running   0          25m
milvus-standalone-milvus-0   1/1     Running   0          24m
milvus-standalone-minio-0    1/1     Running   0          25m

With kbcli installed, you can view comprehensive cluster information:

kbcli cluster describe milvus-standalone -n demo

Name: milvus-standalone	 Created Time: May 19,2025 11:03 UTC+0800
NAMESPACE   CLUSTER-DEFINITION   TOPOLOGY     STATUS    TERMINATION-POLICY
demo        milvus               standalone   Running   Delete

Endpoints:
COMPONENT   INTERNAL   EXTERNAL

Topology:
COMPONENT   SERVICE-VERSION   INSTANCE                     ROLE     STATUS    AZ       NODE    CREATED-TIME
etcd        3.5.15            milvus-standalone-etcd-0     leader   Running   zone-x   x.y.z   May 19,2025 11:03 UTC+0800
milvus      v2.3.2            milvus-standalone-milvus-0   <none>   Running   zone-x   x.y.z   May 19,2025 11:04 UTC+0800
minio       8.0.17            milvus-standalone-minio-0    <none>   Running   zone-x   x.y.z   May 19,2025 11:03 UTC+0800

Resources Allocation:
COMPONENT   INSTANCE-TEMPLATE   CPU(REQUEST/LIMIT)   MEMORY(REQUEST/LIMIT)   STORAGE-SIZE   STORAGE-CLASS
etcd                            500m / 500m          512Mi / 512Mi           data:10Gi      <none>
minio                           500m / 500m          512Mi / 512Mi           data:10Gi      <none>
milvus                          500m / 500m          512Mi / 512Mi           data:10Gi      <none>

Images:
COMPONENT   COMPONENT-DEFINITION              IMAGE
etcd        etcd-3-1.0.0                      quay.io/coreos/etcd:v3.5.15
minio       milvus-minio-1.0.0                docker.io/minio/minio:RELEASE.2022-03-17T06-34-49Z
milvus      milvus-standalone-1.0.0           docker.io/milvusdb/milvus:v2.3.2

Data Protection:
BACKUP-REPO   AUTO-BACKUP   BACKUP-SCHEDULE   BACKUP-METHOD   BACKUP-RETENTION   RECOVERABLE-TIME

Show cluster events: kbcli cluster list-events -n demo milvus-standalone

Access Milvus

To access the Milvus service, you can expose the service by creating a service:

kubectl port-forward pod/milvus-standalone-milvus-0 -n demo 19530:19530

And then you can access the Milvus service via localhost:19530.

Stop the Milvus Cluster

Stopping a cluster temporarily suspends operations while preserving all data and configuration:

Key Effects:

Compute resources (Pods) are released
Persistent storage (PVCs) remains intact
Service definitions are maintained
Cluster configuration is preserved
Operational costs are reduced


apiVersion: operations.kubeblocks.io/v1alpha1
kind: OpsRequest
metadata:
  name: milvus-stop
  namespace: demo
spec:
  clusterName: milvus-standalone
  type: Stop

Alternatively, stop by setting spec.componentSpecs.stop to true:


kubectl patch cluster milvus-standalone -n demo --type='json' -p='[
{
  "op": "add",
  "path": "/spec/componentSpecs/0/stop",
  "value": true
},
{
  "op": "add",
  "path": "/spec/componentSpecs/1/stop",
  "value": true
},
{
  "op": "add",
  "path": "/spec/componentSpecs/2/stop",
  "value": true
}
]'

Start the Milvus Cluster

Restarting a stopped cluster resumes operations with all data and configuration intact.

Key Effects:

Compute resources (Pods) are recreated
Services become available again
Cluster returns to previous state


apiVersion: operations.kubeblocks.io/v1alpha1
kind: OpsRequest
metadata:
  name: milvus-start
  namespace: demo
spec:
  clusterName: milvus-standalone
  type: Start

Restart by setting spec.componentSpecs.stop to false:


kubectl patch cluster milvus-standalone -n demo --type='json' -p='[
{
  "op": "remove",
  "path": "/spec/componentSpecs/0/stop"
},
{
  "op": "remove",
  "path": "/spec/componentSpecs/1/stop"
},
{
  "op": "remove",
  "path": "/spec/componentSpecs/2/stop"
}
]'

Delete Milvus Cluster

Choose carefully based on your data retention needs:

Policy	Resources Removed	Data Removed	Recommended For
DoNotTerminate	None	None	Critical production clusters
Delete	All resources	PVCs deleted	Non-critical environments
WipeOut	All resources	Everything*	Test environments only

*Includes snapshots and backups in external storage

Pre-Deletion Checklist:

Verify no applications are using the cluster
Ensure required backups exist
Confirm proper terminationPolicy is set
Check for dependent resources

For test environments, use this complete cleanup:


kubectl patch cluster milvus-standalone -p '{"spec":{"terminationPolicy":"WipeOut"}}' --type="merge" -n demo
kubectl delete cluster milvus-standalone -n demo