Skip to main content

Deploy JuiceFS S3 Gateway

JuiceFS splits and upload files to the underlying object storage, applications often use the exposed POSIX API. But if you ever need to use S3-compatible API to access JuiceFS files, S3 Gateway comes in handy, its architecture:

JuiceFS S3 Gateway architecture

The feature is implemented based on the MinIO S3 Gateway. It exposes a S3-compatible RESTful API for files on JuiceFS, which is often used in the following scenarios:

  • Expose S3 API for JuiceFS file system, so that applications may access JuiceFS via S3 SDK
  • Use tools like s3cmd, AWS CLI and MinIO Client to access and modify files stored in JuiceFS
  • S3 gateway also provides a file manager that allows users to manage JuiceFS file system directly in web browsers
  • When transferring data across regions, use S3 Gateway as an unified data export endpoint, this eliminates metadata latency and improve performance. See Sync across regions using S3 Gateway

Quick Start​

The S3 gateway can be enabled on the current host using the gateway subcommand of JuiceFS. Before enabling the feature, you need to set the environment variables MINIO_ROOT_USER and MINIO_ROOT_PASSWORD. These are the Access Key and Secret Key for authenticating when accessing the S3 API, and can be simply considered as the username and password of the S3 gateway. For example.

export MINIO_ROOT_USER=admin
export MINIO_ROOT_PASSWORD=12345678
juicefs gateway redis://localhost:6379 localhost:9000

The first two commands of the above three are used to set environment variables. Note that the length of MINIO_ROOT_USER is at least 3 characters, and the length of MINIO_ROOT_PASSWORD is at least 8 characters. If you are a Windows user, replace export with set in the above commands to set the environment variable. i.e., set MINIO_ROOT_USER=admin.

The last command is used to enable the S3 gateway. The gateway subcommand requires at least two parameters. The first is the URL of the database where the metadata is stored, and the second is the address and port on which the S3 gateway is listening. You can add other options to the gateway subcommand to optimize the S3 gateway as needed, for example, to set the default local cache to 20 GiB.

juicefs gateway --cache-size 20480 redis://localhost:6379 localhost:9000

In this example, we assume that the JuiceFS file system is using a local Redis database. When the S3 gateway is enabled, the administrative interface of the S3 gateway can be accessed from the current host using the address http://localhost:9000.

S3-gateway-file-manager

If you want to access the S3 gateway from other hosts on the LAN or over the Internet, you need to change the listening address, e.g.

juicefs gateway redis://localhost:6379 0.0.0.0:9000

In this way, the S3 gateway will accept all network requests by default. S3 clients in different locations can access the S3 gateway using different addresses, e.g.

  • A third-party client in the host where the S3 gateway is located can use http://127.0.0.1:9000 or http://localhost:9000 for access.
  • A third-party client on the same LAN as the host where the S3 gateway is located can access it using http://192.168.1.8:9000 (assuming the intranet IP address of the S3 gateway-enabled host is 192.168.1.8).
  • The S3 gateway can be accessed over the Internet using http://110.220.110.220:9000 (assuming that the public IP address of the S3 gateway-enabled host is 110.220.110.220).

Configure S3 Gateway as Daemon Service​

The S3 gateway can be configured as a daemon service with systemd.

cat > /lib/systemd/system/juicefs-gateway.service<<EOF
[Unit]
Description=Juicefs S3 Gateway
Requires=network.target
After=multi-user.target
StartLimitIntervalSec=0

[Service]
Type=simple
User=root
Environment="MINIO_ROOT_USER=admin"
Environment="MINIO_ROOT_PASSWORD=12345678"
ExecStart=/usr/local/bin/juicefs gateway redis://localhost:6379 localhost:9000
Restart=on-failure
RestartSec=60

[Install]
WantedBy=multi-user.target
EOF

To enable the service at startup

sudo systemctl daemon-reload
sudo systemctl enable juicefs-gateway --now
sudo systemctl status juicefs-gateway

To inspect logs

sudo journalctl -xefu juicefs-gateway.service

Access S3 Gateway​

The JuiceFS S3 gateway can be accessed by various clients, desktop applications, web applications, etc. that support the S3 API. Please note the address and port that the S3 gateway listens on when using it.

tip

The following examples are for using a third-party client to access the S3 gateway running on the local host. In specific scenarios, please adjust the address to access the S3 gateway according to the actual situation.

Use the AWS CLI​

Download and install the AWS CLI from https://aws.amazon.com/cli, then configure:

$ aws configure
AWS Access Key ID [None]: admin
AWS Secret Access Key [None]: 12345678
Default region name [None]:
Default output format [None]:

The program will guide you interactively to add the new configuration, where Access Key ID and Secret Access Key are the same as MINIO_ROOT_USER and MINIO_ROOT_PASSWORD, respectively. The region name and output format should be left blank.

After that, you can access the JuiceFS storage using the aws s3 command, for example:

# List buckets
aws --endpoint-url http://localhost:9000 s3 ls

# List objects in bucket
aws --endpoint-url http://localhost:9000 s3 ls s3://<bucket>

Use the MinIO client​

First install mc by referring to the MinIO download page, then add a new alias:

mc alias set juicefs http://localhost:9000 admin 12345678 --api S3v4

Following the mc command format, the above command creates a configuration with the alias juicefs. Note that the API version -api S3v4 must be specified in the command.

Then, you can freely manage files and folders by copying, moving, adding and deleting between your local disk and JuiceFS storage as well as other cloud storage via the mc client.

$ mc ls juicefs/jfs
[2021-10-20 11:59:00 CST] 130KiB avatar-2191932_1920.png
[2021-10-20 11:59:00 CST] 4.9KiB box-1297327.svg
[2021-10-20 11:59:00 CST]  21KiB cloud-4273197.svg
[2021-10-20 11:59:05 CST]  17KiB hero.svg
[2021-10-20 11:59:06 CST] 1.7MiB hugo-rocha-qFpnvZ_j9HU-unsplash.jpg
[2021-10-20 11:59:06 CST]  16KiB man-1352025.svg
[2021-10-20 11:59:06 CST] 1.3MiB man-1459246.ai
[2021-10-20 11:59:08 CST]  19KiB sign-up-accent-left.07ab168.svg
[2021-10-20 11:59:10 CST]  11MiB work-4997565.svg

Use the virtual-host-style format​

By default, gateway supports path-style requests that are of the format http://mydomain.com/bucket/object. MINIO_DOMAIN environment variable is used to enable virtual-host-style requests. If the request Host header matches with (.+).mydomain.com then the matched pattern $1 is used as bucket and the path is used as object. Example:


export MINIO_DOMAIN=mydomain.com

Deploy JuiceFS S3 Gateway in Kubernetes​

Install via kubectl​

Create a secret (take Amazon S3 as an example):

export NAMESPACE=default
kubectl -n ${NAMESPACE} create secret generic juicefs-secret \
    --from-literal=name=<NAME> \
    --from-literal=metaurl=redis://[:<PASSWORD>]@<HOST>:6379[/<DB>] \
    --from-literal=storage=s3 \
    --from-literal=bucket=https://<BUCKET>.s3.<REGION>.amazonaws.com \
    --from-literal=access-key=<ACCESS_KEY> \
    --from-literal=secret-key=<SECRET_KEY>

Here we have:

  • name: name of the JuiceFS file system.
  • metaurl: URL of the metadata engine (e.g. Redis). Read this document for more information.
  • storage: Object storage type, such as s3, gs, oss. Read this document to find all supported object storages.
  • bucket: Bucket URL. Read this document to learn how to set up different object storage.
  • access-key: Access key of object storage. Read this document for more information.
  • secret-key: Secret key of object storage. Read this document for more information.

Then download the S3 gateway deployment YAML and create the Deployment and Service resources with kubectl. The following points require special attention:

  • Please replace ${NAMESPACE} in the following command with the Kubernetes namespace of the actual S3 gateway deployment, which defaults to kube-system.
  • The replicas for Deployment defaults to 1. Please adjust as needed.
  • The latest version of juicedata/juicefs-csi-driver image is used by default, which has already integrated the latest version of JuiceFS client. Please check here for the specific integrated JuiceFS client version.
  • The initContainers of Deployment will first try to format the JuiceFS file system, if you have already formatted it in advance, this step will not affect the existing JuiceFS file system.
    • The default port number that the S3 gateway listens on is 9000
  • The startup options of S3 gateway will use default values if not specified.
  • The value of MINIO_ROOT_USER environment variable is access-key in Secret, and the value of MINIO_ROOT_PASSWORD environment variable is secret-key in Secret.
curl -sSL https://raw.githubusercontent.com/juicedata/juicefs/main/deploy/juicefs-s3-gateway.yaml | sed "s@kube-system@${NAMESPACE}@g" | kubectl apply -f -

Check if it's deployed successfully:

$ kubectl -n $NAMESPACE get po -o wide -l app.kubernetes.io/name=juicefs-s3-gateway
juicefs-s3-gateway-5c7d65c77f-gj69l         1/1     Running   0          37m     10.244.2.238   kube-node-3   <none>           <none>
$ kubectl -n $NAMESPACE get svc -l app.kubernetes.io/name=juicefs-s3-gateway
NAME                 TYPE        CLUSTER-IP      EXTERNAL-IP   PORT(S)    AGE
juicefs-s3-gateway   ClusterIP   10.101.108.42   <none>        9000/TCP   142m

You can use juicefs-s3-gateway.${NAMESPACE}.svc.cluster.local:9000 or pod IP and port number of juicefs-s3-gateway (e.g. 10.244.2.238:9000) in the application pod to access JuiceFS S3 Gateway.

If you want to access through Ingress, you need to ensure that the Ingress Controller has been deployed in the cluster. Refer to Ingress Controller Deployment Document. Then create an Ingress resource:

kubectl apply -f - <<EOF
apiVersion: networking.k8s.io/v1
kind: Ingress
metadata:
  name: juicefs-s3-gateway
  namespace: ${NAMESPACE}
spec:
  ingressClassName: nginx
  rules:
  - http:
      paths:
      - path: /
        pathType: Prefix
        backend:
          service:
            name: juicefs-s3-gateway
            port:
              number: 9000
EOF

The S3 gateway can be accessed through <external IP> of ingress controller as follows (no need to include the 9000 port number):

kubectl get services -n ingress-nginx

There are some differences between the various versions of Ingress. For more usage methods, please refer to Ingress Controller Usage Document.

Install via Helm​

  1. Prepare a YAML file

    Create a configuration file, for example: values.yaml. Copy and fill in the following configuration information. Among them, the secret part is the information related to the JuiceFS file system, and you can refer to JuiceFS Quick Start Guide for more information.

    values.yaml
    secret:
      name: "<name>"
      metaurl: "<meta-url>"
      storage: "<storage-type>"
      accessKey: "<access-key>"
      secretKey: "<secret-key>"
      bucket: "<bucket>"

    If you want to deploy Ingress, add the following snippet into values.yaml:

    values.yaml
    ingress:
      enabled: true

  2. Deploy

    Execute the following three commands in sequence to deploy the JuiceFS S3 gateway with Helm (note that the following example is deployed to the kube-system namespace).

    helm repo add juicefs-s3-gateway https://juicedata.github.io/charts/
    helm repo update
    helm install juicefs-s3-gateway juicefs-s3-gateway/juicefs-s3-gateway -n kube-system -f ./values.yaml

  3. Check the deployment

    • Check pods are running: the deployment will launch a Deployment named juicefs-s3-gateway, so run kubectl -n kube-system get po -l app.kubernetes.io/name=juicefs-s3-gateway should see all running pods. For example:

      $ kubectl -n kube-system get po -l app.kubernetes.io/name=juicefs-s3-gateway
      NAME                                  READY   STATUS    RESTARTS   AGE
      juicefs-s3-gateway-5c69d574cc-t92b6   1/1     Running   0          136m

    • Check Service: run kubectl -n kube-system get svc -l app.kubernetes.io/name=juicefs-s3-gateway to check Service:

      $ kubectl -n kube-system get svc -l app.kubernetes.io/name=juicefs-s3-gateway
      NAME                 TYPE        CLUSTER-IP      EXTERNAL-IP   PORT(S)    AGE
      juicefs-s3-gateway   ClusterIP   10.101.108.42   <none>        9000/TCP   142m

Monitoring​

Please see the "Monitoring" documentation to learn how to collect and display JuiceFS monitoring metrics.

If you need to use some advanced features of the MinIO S3 gateway, you can pull the gateway branch of this repository and compile MinIO yourself. This branch is developed based on MinIO RELEASE.2022-03-05T06-32-39Z with JuiceFS gateway supported, which supports full functionality of MinIO gateways such as multi-user management while using JuiceFS as a backend.

Compile​

tip

This branch relies on a newer version of JuiceFS. Please refer to the go.mod file for the specific JuiceFS version.

Similar to manually compiling JuiceFS client, you need to install some dependencies in advance to compile S3 gateway.

git clone -b gateway [email protected]:juicedata/minio.git && cd minio
# Will generate a binary named minio
make build

Usage​

The usage of this version of MinIO gateway is exactly the same as that of the native MinIO gateway. For the usage of native functions, please refer to MinIO's document, while JuiceFS's own configuration options can be passed in via the command line. You can use minio gateway juicefs -h to see all currently supported options.

Similar to the S3 gateway integrated with JuiceFS, the gateway service can be started with the following command:

export MINIO_ROOT_USER=admin
export MINIO_ROOT_PASSWORD=12345678
./minio gateway juicefs --console-address ':59001' redis://localhost:6379

The port number of the S3 gateway console is explicitly specified here as 59001. If not specified, a port will be randomly assigned. According to the command line prompt, open the address http://127.0.0.1:59001 in the browser to access the console, as shown in the following snapshot:

S3-gateway-console

Follow us on Twitter
Let's chat on Slack