milvus-logo
LFAI
Home
  • Get Started
  • Benchmarks
Milvus OperatorHelm

Install Milvus Cluster with Milvus Operator

Milvus Operator is a solution that helps you deploy and manage a full Milvus service stack to target Kubernetes (K8s) clusters. The stack includes all Milvus components and relevant dependencies like etcd, Pulsar and MinIO. This topic introduces how to deploy a Milvus cluster with Milvus Operator on K8s.

Prerequisites

Check the requirements for hardware and software prior to your installation.

Create a K8s Cluster

If you have already deployed a K8s cluster for production, you can skip this step and proceed directly to deploy Milvus Operator. If not, you can follow the steps below to quickly create a K8s for testing, and then use it to deploy a Milvus cluster with Milvus Operator.

Create a K8s cluster using minikube

We recommend installing Milvus on K8s with minikube, a tool that allows you to run K8s locally.

minikube can only be used in test environments. It is not recommended that you deploy Milvus distributed clusters in this way in production environments.

1. Install minikube

See install minikube for more information.

2. Start a K8s cluster using minikube

After installing minikube, run the following command to start a K8s cluster.

$ minikube start

3. Check the K8s cluster status

Run $ kubectl cluster-info to check the status of the K8s cluster you just created. Ensure that you can access the K8s cluster via kubectl. If you have not installed kubectl locally, see Use kubectl inside minikube.

Deploy Milvus Operator

Milvus Operator defines a Milvus cluster custom resources on top of Kubernetes Custom Resources. When the custom resources are defined, you can use K8s APIs in a declarative way and manage Milvus deployment stack to ensure its scalability and high-availability.

Prerequisites

  • Ensure that you can access the K8s cluster via kubectl or helm.
  • Ensure the StorageClass dependency is installed as Milvus clusters depend on default StorageClass for data persistence. minikube has a dependency on default StorageClass when installed. Check the dependency by running the command kubectl get sc. If StorageClass is installed, you will see the following output. If not, see Change the Default StorageClass for more information.
NAME                  PROVISIONER                  RECLAIMPOLICY    VOLUMEBIINDINGMODE    ALLOWVOLUMEEXPANSION     AGE
standard (default)    k8s.io/minikube-hostpath     Delete           Immediate             false                    3m36s

1. Install cert-manager

You can install Milvus Operator with Helm or `kubectl` command. If you choose to use Helm, you can skip this step and proceed directly to Install by Helm command.

Milvus Operator uses cert-manager to provide certificate for webhook server. Run the following command to install cert-manager.

$ kubectl apply -f https://github.com/jetstack/cert-manager/releases/download/v1.5.3/cert-manager.yaml

If cert-manager is installed, you can see the following output.

customresourcedefinition.apiextensions.k8s.io/certificaterequests.cert-manager.io created
customresourcedefinition.apiextensions.k8s.io/certificates.cert-manager.io created
customresourcedefinition.apiextensions.k8s.io/challenges.acme.cert-manager.io created
customresourcedefinition.apiextensions.k8s.io/clusterissuers.cert-manager.io created
customresourcedefinition.apiextensions.k8s.io/issuers.cert-manager.io created
customresourcedefinition.apiextensions.k8s.io/orders.acme.cert-manager.io created
namespace/cert-manager created
serviceaccount/cert-manager-cainjector created
serviceaccount/cert-manager created
serviceaccount/cert-manager-webhook created
clusterrole.rbac.authorization.k8s.io/cert-manager-cainjector created
clusterrole.rbac.authorization.k8s.io/cert-manager-controller-issuers created
clusterrole.rbac.authorization.k8s.io/cert-manager-controller-clusterissuers created
clusterrole.rbac.authorization.k8s.io/cert-manager-controller-certificates created
clusterrole.rbac.authorization.k8s.io/cert-manager-controller-orders created
clusterrole.rbac.authorization.k8s.io/cert-manager-controller-challenges created
clusterrole.rbac.authorization.k8s.io/cert-manager-controller-ingress-shim created
clusterrole.rbac.authorization.k8s.io/cert-manager-view created
clusterrole.rbac.authorization.k8s.io/cert-manager-edit created
clusterrole.rbac.authorization.k8s.io/cert-manager-controller-approve:cert-manager-io created
clusterrole.rbac.authorization.k8s.io/cert-manager-controller-certificatesigningrequests created
clusterrole.rbac.authorization.k8s.io/cert-manager-webhook:subjectaccessreviews created
clusterrolebinding.rbac.authorization.k8s.io/cert-manager-cainjector created
clusterrolebinding.rbac.authorization.k8s.io/cert-manager-controller-issuers created
clusterrolebinding.rbac.authorization.k8s.io/cert-manager-controller-clusterissuers created
clusterrolebinding.rbac.authorization.k8s.io/cert-manager-controller-certificates created
clusterrolebinding.rbac.authorization.k8s.io/cert-manager-controller-orders created
clusterrolebinding.rbac.authorization.k8s.io/cert-manager-controller-challenges created
clusterrolebinding.rbac.authorization.k8s.io/cert-manager-controller-ingress-shim created
clusterrolebinding.rbac.authorization.k8s.io/cert-manager-controller-approve:cert-manager-io created
clusterrolebinding.rbac.authorization.k8s.io/cert-manager-controller-certificatesigningrequests created
clusterrolebinding.rbac.authorization.k8s.io/cert-manager-webhook:subjectaccessreviews created
role.rbac.authorization.k8s.io/cert-manager-cainjector:leaderelection created
role.rbac.authorization.k8s.io/cert-manager:leaderelection created
role.rbac.authorization.k8s.io/cert-manager-webhook:dynamic-serving created
rolebinding.rbac.authorization.k8s.io/cert-manager-cainjector:leaderelection created
rolebinding.rbac.authorization.k8s.io/cert-manager:leaderelection created
rolebinding.rbac.authorization.k8s.io/cert-manager-webhook:dynamic-serving created
service/cert-manager created
service/cert-manager-webhook created
deployment.apps/cert-manager-cainjector created
deployment.apps/cert-manager created
deployment.apps/cert-manager-webhook created
mutatingwebhookconfiguration.admissionregistration.k8s.io/cert-manager-webhook created
validatingwebhookconfiguration.admissionregistration.k8s.io/cert-manager-webhook created
cert-manager version 1.1.3 or later is required.

Run $ kubectl get pods -n cert-manager to check if cert-manager is running. You can see the following output if all the pods are running.

NAME                                      READY   STATUS    RESTARTS   AGE
cert-manager-848f547974-gccz8             1/1     Running   0          70s
cert-manager-cainjector-54f4cc6b5-dpj84   1/1     Running   0          70s
cert-manager-webhook-7c9588c76-tqncn      1/1     Running   0          70s

2. Install Milvus Operator

There are two ways to install Milvus Operator on K8s:

  • with helm chart
  • with kubectl command directly with raw manifests

Install by Helm command

helm install milvus-operator \
  -n milvus-operator --create-namespace \
  --wait --wait-for-jobs \
  https://github.com/zilliztech/milvus-operator/releases/download/v0.7.17/milvus-operator-0.7.17.tgz

If Milvus Operator is installed, you can see the following output.

NAME: milvus-operator
LAST DEPLOYED: Thu Jul  7 13:18:40 2022
NAMESPACE: milvus-operator
STATUS: deployed
REVISION: 1
TEST SUITE: None
NOTES:
Milvus Operator Is Starting, use `kubectl get -n milvus-operator deploy/milvus-operator` to check if its successfully installed
If Operator not started successfully, check the checker's log with `kubectl -n milvus-operator logs job/milvus-operator-checker`
Full Installation doc can be found in https://github.com/zilliztech/milvus-operator/blob/main/docs/installation/installation.md
Quick start with `kubectl apply -f https://raw.githubusercontent.com/zilliztech/milvus-operator/main/config/samples/milvus_minimum.yaml`
More samples can be found in https://github.com/zilliztech/milvus-operator/tree/main/config/samples
CRD Documentation can be found in https://github.com/zilliztech/milvus-operator/tree/main/docs/CRD

Install by kubectl command

$ kubectl apply -f https://raw.githubusercontent.com/zilliztech/milvus-operator/main/deploy/manifests/deployment.yaml

If Milvus Operator is installed, you can see the following output.

namespace/milvus-operator created
customresourcedefinition.apiextensions.k8s.io/milvusclusters.milvus.io created
serviceaccount/milvus-operator-controller-manager created
role.rbac.authorization.k8s.io/milvus-operator-leader-election-role created
clusterrole.rbac.authorization.k8s.io/milvus-operator-manager-role created
clusterrole.rbac.authorization.k8s.io/milvus-operator-metrics-reader created
clusterrole.rbac.authorization.k8s.io/milvus-operator-proxy-role created
rolebinding.rbac.authorization.k8s.io/milvus-operator-leader-election-rolebinding created
clusterrolebinding.rbac.authorization.k8s.io/milvus-operator-manager-rolebinding created
clusterrolebinding.rbac.authorization.k8s.io/milvus-operator-proxy-rolebinding created
configmap/milvus-operator-manager-config created
service/milvus-operator-controller-manager-metrics-service created
service/milvus-operator-webhook-service created
deployment.apps/milvus-operator-controller-manager created
certificate.cert-manager.io/milvus-operator-serving-cert created
issuer.cert-manager.io/milvus-operator-selfsigned-issuer created
mutatingwebhookconfiguration.admissionregistration.k8s.io/milvus-operator-mutating-webhook-configuration created
validatingwebhookconfiguration.admissionregistration.k8s.io/milvus-operator-validating-webhook-configuration created

Run $ kubectl get pods -n milvus-operator to check if Milvus Operator is running. You can see the following output if Milvus Operator is running.

NAME                               READY   STATUS    RESTARTS   AGE
milvus-operator-5fd77b87dc-msrk4   1/1     Running   0          46s

Install a Milvus cluster

This tutorial uses the default configuration to install a Milvus cluster. All Milvus cluster components are enabled with multiple replicas, which consumes many resources.

If you have very limited local resources, you can install a Milvus cluster using the minimum configuration).

1. Deploy a Milvus cluster

When Milvus Operator starts, run the following command to deploy a Milvus cluster.

$ kubectl apply -f https://raw.githubusercontent.com/zilliztech/milvus-operator/main/config/samples/milvus_cluster_default.yaml

When the cluster is deployed, you can see the following output.

milvuscluster.milvus.io/my-release created

2. Check the Milvus cluster status

Run the following command to check the status of the Milvus cluster you just deployed.

$ kubectl get milvus my-release -o yaml

You can confirm the current status of Milvus cluster from the status field in the output. When the Milvus cluster is still under creation, the status shows Unhealthy.

apiVersion: milvus.io/v1alpha1
kind: MilvusCluster
metadata:
...
status:
  conditions:
  - lastTransitionTime: "2021-11-02T02:52:04Z"
    message: 'Get "http://my-release-minio.default:9000/minio/admin/v3/info": dial
      tcp 10.96.78.153:9000: connect: connection refused'
    reason: ClientError
    status: "False"
    type: StorageReady
  - lastTransitionTime: "2021-11-02T02:52:04Z"
    message: connection error
    reason: PulsarNotReady
    status: "False"
    type: PulsarReady
  - lastTransitionTime: "2021-11-02T02:52:04Z"
    message: All etcd endpoints are unhealthy
    reason: EtcdNotReady
    status: "False"
    type: EtcdReady
  - lastTransitionTime: "2021-11-02T02:52:04Z"
    message: Milvus Dependencies is not ready
    reason: DependencyNotReady
    status: "False"
    type: MilvusReady
  endpoint: my-release-milvus.default:19530
  status: Unhealthy

Run the following command to check the current status of Milvus pods.

$ kubectl get pods

NAME                                  READY   STATUS              RESTARTS   AGE
my-release-etcd-0                     0/1     Running             0          16s
my-release-etcd-1                     0/1     ContainerCreating   0          16s
my-release-etcd-2                     0/1     ContainerCreating   0          16s
my-release-minio-0                    1/1     Running             0          16s
my-release-minio-1                    1/1     Running             0          16s
my-release-minio-2                    0/1     Running             0          16s
my-release-minio-3                    0/1     ContainerCreating   0          16s
my-release-pulsar-bookie-0            0/1     Pending             0          15s
my-release-pulsar-bookie-1            0/1     Pending             0          15s
my-release-pulsar-bookie-init-h6tfz   0/1     Init:0/1            0          15s
my-release-pulsar-broker-0            0/1     Init:0/2            0          15s
my-release-pulsar-broker-1            0/1     Init:0/2            0          15s
my-release-pulsar-proxy-0             0/1     Init:0/2            0          16s
my-release-pulsar-proxy-1             0/1     Init:0/2            0          15s
my-release-pulsar-pulsar-init-d2t56   0/1     Init:0/2            0          15s
my-release-pulsar-recovery-0          0/1     Init:0/1            0          16s
my-release-pulsar-toolset-0           1/1     Running             0          16s
my-release-pulsar-zookeeper-0         0/1     Pending             0          16s

3. Enable Milvus components

Milvus Operator first creates all dependencies like etcd, Pulsar, and MinIO, and then continues to create Milvus components. Therefore, you can only see the pods of etcd, Pulsar, and MinIO now. Once all denependencies are enabled, Milvus Operator will start all Milvus components. The status of the Milvus cluster is shown as in the following output.

...
status:
  conditions:
  - lastTransitionTime: "2021-11-02T05:59:41Z"
    reason: StorageReady
    status: "True"
    type: StorageReady
  - lastTransitionTime: "2021-11-02T06:06:23Z"
    message: Pulsar is ready
    reason: PulsarReady
    status: "True"
    type: PulsarReady
  - lastTransitionTime: "2021-11-02T05:59:41Z"
    message: Etcd endpoints is healthy
    reason: EtcdReady
    status: "True"
    type: EtcdReady
  - lastTransitionTime: "2021-11-02T06:06:24Z"
    message: '[datacoord datanode indexcoord indexnode proxy querycoord querynode
      rootcoord] not ready'
    reason: MilvusComponentNotHealthy
    status: "False"
    type: MilvusReady
  endpoint: my-release-milvus.default:19530
  status: Unhealthy

Check the status of the Milvus pods again.

$ kubectl get pods

NAME                                            READY   STATUS              RESTARTS   AGE
my-release-etcd-0                               1/1     Running             0          6m49s
my-release-etcd-1                               1/1     Running             0          6m49s
my-release-etcd-2                               1/1     Running             0          6m49s
my-release-milvus-datacoord-6c7bb4b488-k9htl    0/1     ContainerCreating   0          16s
my-release-milvus-datanode-5c686bd65-wxtmf      0/1     ContainerCreating   0          16s
my-release-milvus-indexcoord-586b9f4987-vb7m4   0/1     Running             0          16s
my-release-milvus-indexnode-5b9787b54-xclbx     0/1     ContainerCreating   0          16s
my-release-milvus-proxy-84f67cdb7f-pg6wf        0/1     ContainerCreating   0          16s
my-release-milvus-querycoord-865cc56fb4-w2jmn   0/1     Running             0          16s
my-release-milvus-querynode-5bcb59f6-nhqqw      0/1     ContainerCreating   0          16s
my-release-milvus-rootcoord-fdcccfc84-9964g     0/1     Running             0          16s
my-release-minio-0                              1/1     Running             0          6m49s
my-release-minio-1                              1/1     Running             0          6m49s
my-release-minio-2                              1/1     Running             0          6m49s
my-release-minio-3                              1/1     Running             0          6m49s
my-release-pulsar-bookie-0                      1/1     Running             0          6m48s
my-release-pulsar-bookie-1                      1/1     Running             0          6m48s
my-release-pulsar-bookie-init-h6tfz             0/1     Completed           0          6m48s
my-release-pulsar-broker-0                      1/1     Running             0          6m48s
my-release-pulsar-broker-1                      1/1     Running             0          6m48s
my-release-pulsar-proxy-0                       1/1     Running             0          6m49s
my-release-pulsar-proxy-1                       1/1     Running             0          6m48s
my-release-pulsar-pulsar-init-d2t56             0/1     Completed           0          6m48s
my-release-pulsar-recovery-0                    1/1     Running             0          6m49s
my-release-pulsar-toolset-0                     1/1     Running             0          6m49s
my-release-pulsar-zookeeper-0                   1/1     Running             0          6m49s
my-release-pulsar-zookeeper-1                   1/1     Running             0          6m
my-release-pulsar-zookeeper-2                   1/1     Running             0          6m26s

When all components are enabled, the status of the Milvus cluster is shown as Healthy.

...
status:
  conditions:
  - lastTransitionTime: "2021-11-02T05:59:41Z"
    reason: StorageReady
    status: "True"
    type: StorageReady
  - lastTransitionTime: "2021-11-02T06:06:23Z"
    message: Pulsar is ready
    reason: PulsarReady
    status: "True"
    type: PulsarReady
  - lastTransitionTime: "2021-11-02T05:59:41Z"
    message: Etcd endpoints is healthy
    reason: EtcdReady
    status: "True"
    type: EtcdReady
  - lastTransitionTime: "2021-11-02T06:12:36Z"
    message: All Milvus components are healthy
    reason: MilvusClusterHealthy
    status: "True"
    type: MilvusReady
  endpoint: my-release-milvus.default:19530
  status: Healthy

Check the status of the Milvus pods again. You can see all the pods are running now.

$ kubectl get pods
NAME                                            READY   STATUS      RESTARTS   AGE
my-release-etcd-0                               1/1     Running     0          14m
my-release-etcd-1                               1/1     Running     0          14m
my-release-etcd-2                               1/1     Running     0          14m
my-release-milvus-datacoord-6c7bb4b488-k9htl    1/1     Running     0          6m
my-release-milvus-datanode-5c686bd65-wxtmf      1/1     Running     0          6m
my-release-milvus-indexcoord-586b9f4987-vb7m4   1/1     Running     0          6m
my-release-milvus-indexnode-5b9787b54-xclbx     1/1     Running     0          6m
my-release-milvus-proxy-84f67cdb7f-pg6wf        1/1     Running     0          6m
my-release-milvus-querycoord-865cc56fb4-w2jmn   1/1     Running     0          6m
my-release-milvus-querynode-5bcb59f6-nhqqw      1/1     Running     0          6m
my-release-milvus-rootcoord-fdcccfc84-9964g     1/1     Running     0          6m
my-release-minio-0                              1/1     Running     0          14m
my-release-minio-1                              1/1     Running     0          14m
my-release-minio-2                              1/1     Running     0          14m
my-release-minio-3                              1/1     Running     0          14m
my-release-pulsar-bookie-0                      1/1     Running     0          14m
my-release-pulsar-bookie-1                      1/1     Running     0          14m
my-release-pulsar-bookie-init-h6tfz             0/1     Completed   0          14m
my-release-pulsar-broker-0                      1/1     Running     0          14m
my-release-pulsar-broker-1                      1/1     Running     0          14m
my-release-pulsar-proxy-0                       1/1     Running     0          14m
my-release-pulsar-proxy-1                       1/1     Running     0          14m
my-release-pulsar-pulsar-init-d2t56             0/1     Completed   0          14m
my-release-pulsar-recovery-0                    1/1     Running     0          14m
my-release-pulsar-toolset-0                     1/1     Running     0          14m
my-release-pulsar-zookeeper-0                   1/1     Running     0          14m
my-release-pulsar-zookeeper-1                   1/1     Running     0          13m
my-release-pulsar-zookeeper-2                   1/1     Running     0          13m

When the Milvus cluster is installed, you can learn how to Connect to Milvus server.

Connect to Milvus

Verify which local port the Milvus server is listening on. Replace the pod name with your own.

$ kubectl get pod my-release-milvus-proxy-84f67cdb7f-pg6wf --template
='{{(index (index .spec.containers 0).ports 0).containerPort}}{{"\n"}}'
19530

Open a new terminal and run the following command to forward a local port to the port that Milvus uses. Optionally, omit the designated port and use :19530 to let kubectl allocate a local port for you so that you don’t have to manage port conflicts.

$ kubectl port-forward service/my-release-milvus 27017:19530
Forwarding from 127.0.0.1:27017 -> 19530

By default, kubectl’s port-forwarding only listens on localhost. Use flag address if you want Milvus server to listen on selected IP or all addresses.

$ kubectl port-forward --address 0.0.0.0 service/my-release-milvus 27017:19530
Forwarding from 0.0.0.0:27017 -> 19530

Uninstall the Milvus cluster

Run the following command to uninstall the Milvus cluster.

$ kubectl delete milvus my-release
  • When you delete the Milvus cluster using the default configuration, dependencies like etcd, Pulsar, and MinIO are not deleted. Therefore, next time when you install the same Milvus cluster instance, these dependencies will be used again.
  • To delete the dependencies and private virtual clouds (PVCs) along with the Milvus cluster, see configuration file.

    • Uninstall Milvus Operator

    There are also two ways to uninstall Milvus Operator on K8s:

    Uninstall Milvus Operator by Helm command

    $ helm -n milvus-operator uninstall milvus-operator

    Uninstall Milvus Operator by kubectl command

    $ kubectl delete -f https://raw.githubusercontent.com/zilliztech/milvus-operator/v0.7.17/deploy/manifests/deployment.yaml

    Delete the K8s cluster

    When you no longer need the K8s cluster in the test environment, you can run $ minikube delete to delete it.

    What’s next

    Having installed Milvus, you can:

    Table of contents

    Try Managed Milvus for Free

    Zilliz Cloud is hassle-free, powered by Milvus and 10x faster.

    Get Started
    Feedback

    Was this page helpful?