### Pip download output example

Source: https://d7y.io/docs/operations/integrations/pip

This is an example of the expected output when pip downloads a package through the Dragonfly proxy, showing the collection and installation process.

```text
Looking in indexes: https://pypi.org/simple/
Collecting torch
  Using cached torch-2.5.1-cp311-cp311-manylinux2014_aarch64.whl (91.9 MB)
Collecting filelock
  Using cached filelock-3.17.0-py3-none-any.whl (16 kB)
Collecting typing-extensions>=4.8.0
  Downloading typing_extensions-4.12.2-py3-none-any.whl (37 kB)
Collecting networkx
  Downloading networkx-3.4.2-py3-none-any.whl (1.7 MB)
     ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 1.7/1.7 MB 195.8 MB/s eta 0:00:00
Collecting jinja2
  Downloading jinja2-3.1.5-py3-none-any.whl (134 kB)
     ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 134.6/134.6 kB 38.5 MB/s eta 0:00:00
Collecting fsspec
  Downloading fsspec-2024.12.0-py3-none-any.whl (183 kB)
     ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 183.9/183.9 kB 37.5 MB/s eta 0:00:00
Collecting sympy==1.13.1
  Downloading sympy-1.13.1-py3-none-any.whl (6.2 MB)
     ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 6.2/6.2 MB 245.3 MB/s eta 0:00:00
Collecting mpmath<1.4,>=1.1.0
  Downloading mpmath-1.3.0-py3-none-any.whl (536 kB)
     ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 536.2/536.2 kB 170.5 MB/s eta 0:00:00
Collecting MarkupSafe>=2.0
  Downloading MarkupSafe-3.0.2-cp311-cp311-manylinux_2_17_aarch64.manylinux2014_aarch64.whl (23 kB)
Installing collected packages: mpmath, typing-extensions, sympy, networkx, MarkupSafe, fsspec, filelock, jinja2, torch
Successfully installed MarkupSafe-3.0.2 filelock-3.17.0 fsspec-2024.12.0 jinja2-3.1.5 mpmath-1.3.0 networkx-3.4.2 sympy-1.13.1 torch-2.5.1 typing-extensions-4.12.2
  

```

--------------------------------

### Start Minikube with Podman

Source: https://d7y.io/docs/operations/integrations/container-runtime/podman

Use this command to start a Minikube cluster with Podman as the container runtime.

```bash
minikube start --driver=podman --container-runtime=cri-o  

```

--------------------------------

### Start Minikube with CRI-O

Source: https://d7y.io/docs/operations/integrations/container-runtime/cri-o

Use this command to start a Minikube cluster with CRI-O as the container runtime. Ensure CRI-O is installed and configured on your system.

```bash
minikube start --container-runtime=cri-o
```

--------------------------------

### Run Scheduler CLI

Source: https://d7y.io/docs/getting-started/installation/binaries

View Scheduler CLI help documentation or start the Scheduler service.

```bash
# View Scheduler cli help docs.  
scheduler --help  
  
# Setup Scheduler, it is recommended to start Scheduler via systemd.  
scheduler  

```

--------------------------------

### Run Manager CLI

Source: https://d7y.io/docs/getting-started/installation/binaries

View Manager CLI help documentation or start the Manager service.

```bash
# View Manager cli help docs.  
manager --help  
  
# Setup Manager, it is recommended to start Manager via systemd.  
manager  

```

--------------------------------

### Install Python package with pip

Source: https://d7y.io/docs/operations/integrations/pip

Use the `pip install` command to download and install a Python package, such as `torch`. Pip will now use the configured Dragonfly proxy.

```bash
pip install torch

```

--------------------------------

### Install Dragonfly with Helm

Source: https://d7y.io/docs/operations/integrations/container-runtime/podman

Install the Dragonfly cluster using Helm charts and the provided values.yaml configuration file.

```bash
$ helm repo add dragonfly https://dragonflyoss.github.io/helm-charts/  
$ helm install --create-namespace --namespace dragonfly-system dragonfly dragonfly/dragonfly -f values.yaml  

```

--------------------------------

### Install Dragonfly with Helm

Source: https://d7y.io/docs/getting-started/installation/helm-charts

Add the Dragonfly Helm repository and install the Dragonfly chart using the prepared values.yaml file. This command creates the dragonfly-system namespace if it doesn't exist.

```bash
helm repo add dragonfly https://dragonflyoss.github.io/helm-charts/
helm install --create-namespace --namespace dragonfly-system dragonfly dragonfly/dragonfly -f values.yaml
```

--------------------------------

### Manager YAML Configuration Example

Source: https://d7y.io/docs/reference/configuration/manager

This snippet shows a comprehensive example of the manager.yaml configuration file. It includes settings for gRPC and REST servers, logging, authentication, and database connections. Use this as a template for your own manager configuration.

```yaml
# Current server info used for server.  

server:  

  # GRPC server configure.  

  grpc:  

    # # Access ip for other services,  

    # # when local ip is different with access ip, advertiseIP should be set.  

    # advertiseIP: 127.0.0.1  

    # # Listen ip.  

    # listenIP: 0.0.0.0  

    # Listen port.  

    # when this port is not available, manager will try next port.  

    port:  

      start: 65003  

      end: 65003  

    # RequestRateLimit is the maximum number of requests per second for the gRPC server. It limits both the rate of  

    # unary gRPC requests and the rate of new stream gRPC connection, default is 4000 req/s.  

    requestRateLimit: 4000  

  

  # # GRPC server tls configuration.  

  # tls:  

  #   # CA certificate file path for mTLS.  

  #   caCert: /etc/ssl/certs/ca.crt  

  #   # Certificate file path for mTLS.  

  #   cert: /etc/ssl/certs/server.crt  

  #   # Key file path for mTLS.  

  #   key: /etc/ssl/private/server.pem  

  

  # REST server configure  

  rest:  

    # REST server address  

    addr: :8080  

  # # REST server tls configuration.  

  # tls:  

  #   # Certificate file path.  

  #   cert: /etc/ssl/certs/server.crt  

  #   # Key file path.  

  #   key: /etc/ssl/private/server.pem  

  

  # logLevel specifies the logging level for the manager.  

  # Default: "info"  

  # Supported values: "debug", "info", "warn", "error", "panic", "fatal"  

  logLevel: "info"  

  

  # logMaxSize specifies maximum size in megabytes of log files before rotation for the manager.  

  # Default: 1024  

  logMaxSize: 1024  

  

  # logMaxAge specifies maximum number of days to retain old log files for the manager.  

  # Default: 7  

  logMaxAge: 7  

  

  # logMaxBackups specifies maximum number of old log files to keep for the manager.  

  # Default: 20  

  logMaxBackups: 20  

  

  # logDir is the log directory.  

  # In linux, default value is /var/log/dragonfly.  

  # In macos(just for testing), default value is /Users/$USER/.dragonfly/logs.  

  logDir: ''  

  

  # cacheDir is dynconfig cache directory.  

  # In linux, default value is /var/cache/dragonfly.  

  # In macos(just for testing), default value is /Users/$USER/.dragonfly/cache.  

  cacheDir: ''  

  

  # pluginDir is the plugin directory.  

  # In linux, default value is /usr/local/dragonfly/plugins.  

  # In macos(just for testing), default value is /Users/$USER/.dragonfly/plugins.  

  pluginDir: ''  

  # dataDir is the data directory.  

  # In linux, default value is /usr/local/dragonfly/data.  

  # In macos(just for testing), default value is /Users/$USER/.dragonfly/data.  

  dataDir: ''  

  

# Auth configuration.  

auth:  

  # JWT configuration used for signing.  

  jwt:  

    # Realm name to display to the user, default value is Dragonfly.  

    realm: 'Dragonfly'  

    # Key is secret key used for signing, default value is  

    # encoded base64 of dragonfly.  

    # Please change the key in production.  

    key: 'ZHJhZ29uZmx5Cg=='  

    # Timeout is duration that a jwt token is valid,  

    # default duration is two days.  

    timeout: 48h  

    # MaxRefresh field allows clients to refresh their token  

    # until MaxRefresh has passed, default duration is two days.  

    maxRefresh: 48h  

  

# Database info used for server.  

database:  

  # Database type, supported types include mysql, mariadb and postgres.  

  type: mysql  

  # Mysql configure.  

  mysql:  

    user: dragonfly  

    password: dragonfly  

    host: dragonfly  

    port: 3306  

    dbname: manager  

    migrate: true  

  # Postgres configure.  

  postgres:  

    user: dragonfly  

    password: dragonfly  

    host: dragonfly  

    port: 5432  

    dbname: manager  

    sslMode: disable  

    timezone: UTC  

    migrate: true  

  # tlsConfig: preferred  

  # tls:  

  #   # Client certificate file path.  

  #   cert: /etc/ssl/certs/cert.pem  

  #   # Client key file path.  

  #   key: /etc/ssl/private/key.pem  

  #   # CA file path.  

  #   ca: /etc/ssl/certs/ca.pem  

  #   # Whether a client verifies the server's certificate chain and hostname.  

  #   insecureSkipVerify: true  

  # Redis configure.  

  redis:  

    # Redis addresses.  

    addrs:  

      - dragonfly:6379  

    # Redis sentinel master name.  

    masterName: ''  

    # Redis username.  

    username: ''  

    # Redis password.  

    password: ''  

    # Redis DB.  

    db: 0  

    # Redis broker DB.  

    brokerDB: 1  

    # Redis backend DB.  

    backendDB: 2  

  # # Redis TLS client configuration.  

  # tls:  

  
```

--------------------------------

### Run Dfdaemon as Seed Peer CLI

Source: https://d7y.io/docs/getting-started/installation/binaries

View Dfdaemon CLI help documentation or start Dfdaemon as a Seed Peer.

```bash
# View Dfdaemon cli help docs.  
dfdaemon --help  
  
# Setup Dfdaemon, it is recommended to start Dfdaemon via systemd.  
dfdaemon  

```

--------------------------------

### Git LFS Download Log Example

Source: https://d7y.io/docs/operations/integrations/git-lfs

An example of verbose Git LFS logs showing download actions and the content storage URL. Look for 'trace git-lfs' and details within 'actions.download.href'.

```log
18:52:51.137490 trace git-lfs: HTTP: {"objects":[{"oid":"68ac0af011ce9c51a4c74c5ac9a40218e9e67bf55ebe13c8f2d758f710a3163a","size":19670194,"actions":{"download":{"href":"https://github-cloud.githubusercontent.com/alambic/media/730487717/68/ac/68ac0af011ce9c51a4c74c5ac9a40218e9e67bf55ebe13c8f2d758f710a3163a?X-Amz-Algorithm=AWS4-HMAC-SHA256&X-Amz-Credential=AKIA5BA2674WPWWEFGQ5%2F20240605%2Fus-east-1%2Fs3%2Faws4_request&X-Amz-Date=20240605T105251Z&X-Amz-Expires=3600&X-Amz-Signature=35cf8e02f0d3e2da893aa46fa4929d79ce1abb18aea8e0fabfbb138706d7151818:52:51.137574 trace git-lfs: HTTP: &X-Amz-SignedHeaders=host&actor_id=0&key_id=0&repo_id=810214636&token=1"}}}]}
```

--------------------------------

### Manage dfdaemon Service with Systemd

Source: https://d7y.io/docs/getting-started/installation/binaries

Commands to enable, start, and check the status of the dfdaemon service using systemd. Ensure the manager and scheduler are running before starting dfdaemon.

```bash
$ sudo systemctl enable dfdaemon
$ sudo systemctl start dfdaemon
$ sudo systemctl status dfdaemon
```

--------------------------------

### Install Prometheus and Grafana Stack

Source: https://d7y.io/docs/operations/best-practices/observability/monitoring

Installs the kube-prometheus-stack Helm chart, which includes Prometheus and Grafana, into a specified namespace. It uses a custom values file for configuration.

```bash
$ helm install --create-namespace --namespace prometheus prometheus prometheus-community/kube-prometheus-stack -f https://raw.githubusercontent.com/dragonflyoss/monitoring/main/prometheus/values.yaml
```

--------------------------------

### Example Triton Model Configuration with Dragonfly Agent

Source: https://d7y.io/docs/operations/integrations/triton-server

An example of a `config.pbtxt` file for a model, including the Dragonfly agent integration. This configuration specifies model details and enables Dragonfly for repository access.

```protobuf
name: "densenet_onnx"  
platform: "onnxruntime_onnx"  
max_batch_size : 0  
input [
  {
    name: "data_0"
    data_type: TYPE_FP32
    format: FORMAT_NCHW
    dims: [ 3, 224, 224 ]
    reshape { shape: [ 1, 3, 224, 224 ] }
  }
]
output [
  {
    name: "fc6_1"
    data_type: TYPE_FP32
    dims: [ 1000 ]
    reshape { shape: [ 1, 1000, 1, 1 ] }
    label_filename: "densenet_labels.txt"
  }
]
model_repository_agents  
{
  agents [
    {
      name: "dragonfly",
    }
  ]
}

```

--------------------------------

### Run Dfdaemon with Systemd

Source: https://d7y.io/docs/getting-started/installation/binaries

Enable and start the dfdaemon service using systemctl. This assumes the dfdaemon executable and configuration are in place. The service status can be checked afterwards.

```bash
$ sudo systemctl enable dfdaemon  
$ sudo systemctl start dfdaemon  
$ sudo systemctl status dfdaemon  
● dfdaemon.service - dfdaemon is a high performance P2P download daemon  
     Loaded: loaded (/lib/systemd/system/dfdaemon.service; enabled; preset: enabled)  
     Active: active (running) since Mon 2024-08-05 17:46:39 UTC; 4s ago  
       Docs: https://github.com/dragonflyoss/client  
   Main PID: 2118 (dfdaemon)  
      Tasks: 13 (limit: 11017)  
     Memory: 15.0M (max: 8.0G available: 7.9G)  
        CPU: 83ms  
     CGroup: /system.slice/dfdaemon.service  
             └─2118 /usr/bin/dfdaemon --config /etc/dragonfly/dfdaemon.yaml --console  
```

--------------------------------

### Download File with dfget

Source: https://d7y.io/docs/concepts/task

Use dfget to download files from a specified URL. Ensure the dfget utility is installed and accessible.

```bash
$ dfget https://<host>:<port>/<path> -O /tmp/file.txt  

[00:00:00] [============================================================] 100% (209.63 KiB/s, 0.0s)  

```

--------------------------------

### Extract and Install Nydus Snapshotter

Source: https://d7y.io/docs/operations/integrations/container-runtime/nydus

Extract the downloaded Nydus Snapshotter archive and install the `containerd-nydus-grpc` executable to `/usr/local/bin/`.

```bash
tar zxvf nydus-snapshotter_linux_arm64.tar.gz  

# Install executable file to  /usr/local/bin/{containerd-nydus-grpc}.
sudo cp bin/containerd-nydus-grpc /usr/local/bin/  
```

--------------------------------

### Run Unit Tests

Source: https://d7y.io/docs/development-guide/running-tests

Execute all unit tests in the project. Ensure Golang and Rust are installed and meet the version requirements.

```bash
make test
```

--------------------------------

### Download and Install Dragonfly Client DEB Package

Source: https://d7y.io/docs/getting-started/installation/binaries

Download the Dragonfly client DEB package using wget and install it using dpkg. Replace `{arch}` with your system's architecture (e.g., x86_64, aarch64) and `{version}` with the desired client version.

```bash
wget -O dragonfly-client-{arch}-unknown-linux-musl.deb \
  https://github.com/dragonflyoss/client/releases/download/v{version}/dragonfly-client-v{version}-{arch}-unknown-linux-musl.deb

dpkg -i dragonfly-client-{arch}-unknown-linux-musl.deb
```

--------------------------------

### Start TorchServe with Dragonfly Endpoint Plugin

Source: https://d7y.io/docs/operations/integrations/torchserve

Start TorchServe, specifying the model store and plugins path. Ensure the plugins-path points to the directory containing the Dragonfly Endpoint JAR file.

```bash
torchserve --start --model-store <path-to-model-store-file> --plugins-path=<path-to-plugin-jars>
```

--------------------------------

### List All Tasks

Source: https://d7y.io/docs/reference/commands/client/dfctl

Lists all tasks currently stored in the client's local storage. Use this command to get an overview of available tasks.

```bash
dfctl task ls
```

--------------------------------

### Dfdaemon Peer Configuration

Source: https://d7y.io/docs/getting-started/installation/binaries

Example configuration for setting up dfdaemon as a regular peer. This configuration only requires the manager address.

```yaml
# Peer configuration.  

manager:  
  addr: http://dragonfly-manager:65003  
```

--------------------------------

### Install Dragonfly using Helm

Source: https://d7y.io/docs/getting-started/quick-start/kubernetes

Add the Dragonfly Helm repository and install the Dragonfly cluster using the specified configuration file. The '--wait' flag ensures the command waits for the deployment to complete.

```bash
$ helm repo add dragonfly https://dragonflyoss.github.io/helm-charts/
$ helm install --wait --create-namespace --namespace dragonfly-system dragonfly dragonfly/dragonfly -f charts-config.yaml

```

--------------------------------

### Install Dragonfly using Helm

Source: https://d7y.io/docs/getting-started/installation/helm-charts

Add the Dragonfly Helm repository and install the Dragonfly cluster using the `values.yaml` configuration file. The `--wait` flag ensures the command waits for resources to be ready.

```bash
$ helm repo add dragonfly https://dragonflyoss.github.io/helm-charts/
$ helm install --wait --create-namespace --namespace dragonfly-system dragonfly dragonfly/dragonfly -f values.yaml

```

--------------------------------

### Install Dragonfly Client using RPM

Source: https://d7y.io/docs/getting-started/installation/binaries

Download the Dragonfly client RPM package and install it using the 'rpm -ivh' command. Ensure to replace '{arch}' with the correct architecture (e.g., x86_64) and '{version}' with the desired version.

```bash
wget -O dragonfly-client-{arch}-unknown-linux-musl.rpm https://github.com/dragonflyoss/client/releases/download/v{version}/dragonfly-client-v{version}-{arch}-unknown-linux-musl.rpm  


rpm -ivh dragonfly-client-{arch}-unknown-linux-musl.rpm  
```

--------------------------------

### Dragonfly download log output example

Source: https://d7y.io/docs/operations/integrations/pip

This is an example of the log output from a Dragonfly client pod, indicating a successful download task. It includes timestamps, host IDs, and task IDs.

```json
{
  2025-01-23T05:39:15.828741839+00:00  INFO  
  download_task: dragonfly-client/src/grpc/dfdaemon_upload.rs:395: download task succeeded  
  host_id="10.244.2.15-dragonfly-seed-client-2-seed"  
  task_id="92d8d9e20c608edcca38790dfc76b99746e65592321923b552afe6337b275257"  
  peer_id="10.244.2.15-dragonfly-seed-client-2-01f2e222-f517-47d8-98b4-18540e4c5be5-seed"  
}
{
  2025-01-23T05:39:18.955058382+00:00  
  INFO download_task: dragonfly-client/src/grpc/dfdaemon_upload.rs:395: download task succeeded  
  host_id="10.244.2.15-dragonfly-seed-client-2-seed"  
  task_id="7f684c300a629072a61536344a0369df2e049829ab8ab703e6de145cef18eb14"  
  peer_id="10.244.2.15-dragonfly-seed-client-2-0f27a38b-97af-45c7-9e57-43cbc2bcd8d9-seed"  
}

```

--------------------------------

### Install Dragonfly Cluster with Helm

Source: https://d7y.io/docs/operations/integrations/container-runtime/nydus

Commands to add the Dragonfly Helm repository and install a Dragonfly cluster using a custom configuration file. Includes options for waiting for deployment and creating namespaces.

```bash
$ helm repo add dragonfly https://dragonflyoss.github.io/helm-charts/
$ helm install --wait --create-namespace --namespace dragonfly-system dragonfly dragonfly/dragonfly -f charts-config.yaml
```

--------------------------------

### Triton Inference Server Image Client Example

Source: https://d7y.io/docs/operations/integrations/triton-server

Use this Docker command to run the Triton image client and perform inference on an image. This verifies the inference API is responding correctly.

```bash
docker run -it --rm --net=host nvcr.io/nvidia/tritonserver:23.08-py3-sdk /workspace/install/bin/image_client -m densenet_onnx -c 3 -s INCEPTION /workspace/images/mug.jpg  
```

--------------------------------

### Containerd Helm Chart Configuration

Source: https://d7y.io/docs/operations/integrations/container-runtime/containerd

Example values.yaml for deploying containerd via Helm charts, specifying image repositories, metrics enablement, and private registry configurations.

```yaml
manager:  
  image:  
    repository: dragonflyoss/manager  
    tag: latest  
  metrics:  
    enable: true  
  
scheduler:  
  image:  
    repository: dragonflyoss/scheduler  
    tag: latest  
  metrics:  
    enable: true  
  
seedClient:  
  image:  
    repository: dragonflyoss/client  
    tag: latest  
  metrics:  
    enable: true  
  
client:  
  image:  
    repository: dragonflyoss/client  
    tag: latest  
  metrics:  
    enable: true  
  dfinit:  
    enable: true  
    image:  
      repository: dragonflyoss/dfinit  
      tag: latest  
    config:  
      containerRuntime:  
        containerd:  
          configPath: /etc/containerd/config.toml  
          registries:  
            - hostNamespace: your_private_registry_host_addr  
              serverAddr: your_private_registry_server_addr  
              capabilities: ['pull', 'resolve']  

```

--------------------------------

### Use Dfget to Download Files

Source: https://d7y.io/docs/getting-started/installation/binaries

Demonstrates how to use the dfget command-line tool to download files. Includes viewing help documentation and downloading a file via HTTP protocol.

```bash
# View Dfget cli help docs.  

dfget --help  

# Download with HTTP protocol  
dfget -O /path/to/output http://example.com/object  
```

--------------------------------

### Create and Prepare Model Store Directory

Source: https://d7y.io/docs/operations/integrations/torchserve

Create a directory to store model files and set appropriate permissions for it.

```bash
mkdir model-store
chmod 777 model-store
```

--------------------------------

### Enable and Start Nydus Snapshotter Service

Source: https://d7y.io/docs/operations/integrations/container-runtime/nydus

Commands to enable the Nydus snapshotter service to start on boot, start the service immediately, and check its status.

```bash
$ sudo systemctl enable nydus-snapshotter
$ sudo systemctl start nydus-snapshotter
$ sudo systemctl status nydus-snapshotter
```

--------------------------------

### Verify Dragonfly Binaries Availability

Source: https://d7y.io/docs/getting-started/installation/helm-charts

Execute `kubectl exec` to enter the pod and use the `which` command to confirm that the Dragonfly binaries (dfget, dfcache, dfstore, dfdaemon) are installed and accessible within the container's PATH.

```bash
$ kubectl exec -it test-pod -- bash
$ which dfget dfcache dfstore dfdaemon
```

--------------------------------

### View Dfget Help and Download Files

Source: https://d7y.io/docs/getting-started/installation/binaries

View the help documentation for the dfget command and use it to download files via HTTP. Specify the output path with `-O`.

```bash
# View Dfget cli help docs. 
dfget --help

# Download with HTTP protocol 
dfget -O /path/to/output http://example.com/object
```

--------------------------------

### Install Dragonfly using Helm

Source: https://d7y.io/docs/reference/commands/client/dfcache

Install Dragonfly components into a Kubernetes cluster using Helm. This command adds the Dragonfly Helm repository and installs the Dragonfly chart.

```bash
helm repo add dragonfly https://dragonflyoss.github.io/helm-charts/
```

```bash
helm install --create-namespace -n dragonfly-system dragonfly dragonfly/dragonfly
```

--------------------------------

### Install cert-manager

Source: https://d7y.io/docs/getting-started/installation/helm-charts

Install cert-manager, a prerequisite for the Dragonfly webhook injector, using its official manifest.

```bash
kubectl apply -f https://github.com/cert-manager/cert-manager/releases/latest/download/cert-manager.yaml
```

--------------------------------

### Download a Single File with Authentication

Source: https://d7y.io/docs/operations/integrations/model-scope

To access private repositories or increase rate limits, use the --ms-token flag with your authentication token.

```bash
dfget modelscope://deepseek-ai/DeepSeek-R1/config.json -O /tmp/config.json --ms-token=<token>
```

--------------------------------

### Install TorchServe Dependencies and Model Archiver

Source: https://d7y.io/docs/operations/integrations/torchserve

Install necessary Python dependencies and the TorchServe model archiver using pip and conda.

```bash
python ./ts_scripts/install_dependencies.py  
```

```bash
conda install torchserve torch-model-archiver torch-workflow-archiver -c pytorch  
```

--------------------------------

### Extract and Install Nydus Image Service

Source: https://d7y.io/docs/operations/integrations/container-runtime/nydus

Extract the Nydus Image Service archive and install the `nydus-image`, `nydusd`, and `nydusify` executables to `/usr/local/bin/`.

```bash
tar zxvf nydus-image-service-linux-arm64.tgz  

# Install executable file to  /usr/local/bin/{nydus-image,nydusd,nydusify}.  
sudo cp nydus-static/nydus-image nydus-static/nydusd nydus-static/nydusify /usr/local/bin/  
```

--------------------------------

### Dfget Basic Usage

Source: https://d7y.io/docs/reference/commands/client/dfget

Shows the basic syntax for using dfget to download files. The output path and URL are required parameters.

```bash
dfget -O <OUTPUT> <URL>
```

```bash
dfget [command]
```

--------------------------------

### Get Task Status by Job ID

Source: https://d7y.io/docs/advanced-guides/open-api/task

Retrieves the status of a previously created task using its job ID. This GET request is used for polling.

```APIDOC
## GET /oapi/v1/jobs/{job_id}

### Description
Polls the status of a specific job using its unique identifier.

### Method
GET

### Endpoint
http://dragonfly-manager:8080/oapi/v1/jobs/{job_id}

### Parameters
#### Path Parameters
- **job_id** (integer) - Required - The ID of the job to retrieve the status for.

### Response
#### Success Response (200)
- **id** (integer) - The unique identifier for the job.
- **created_at** (string) - Timestamp when the job was created.
- **updated_at** (string) - Timestamp when the job was last updated.
- **type** (string) - The type of task.
- **state** (string) - The current state of the task (e.g., "SUCCESS").
- **args** (object) - Arguments used for the task.
- **result** (object) - The result of the task execution, if available.
  - **job_states** (array) - An array containing job state details.
    - **results** (array) - An array of results, which may include peer addresses.
      - **peers** (array) - A list of peer addresses if the task was successful.
  - **state** (string) - The overall state of the result.
  - **updated_at** (string) - Timestamp when the result was last updated.
- **scheduler_clusters** (array) - List of scheduler clusters associated with the job.

#### Response Example
```json
{
    "id": 1,
    "created_at": "2024-11-19T08:08:23Z",
    "updated_at": "2024-11-19T08:08:53Z",
    "type": "get_task",
    "state": "SUCCESS",
    "args": {
        "content_for_calculating_task_id": "your_content_for_calculating_task_id"
    },
    "result": {
        "job_states": [
            {
                "results": [
                    {
                        "peers": [
                            {
                                "created_at": "2024-11-19T08:13:35.210473555Z",
                                "host_type": "super",
                                "hostname": "dragonfly-seed-client-0",
                                "id": "10.244.1.32-dragonfly-seed-client-0-9783271e-e27c-4b16-a7c9-6a8ffe18cd1a-seed",
                                "ip": "10.244.1.32",
                                "updated_at": "2024-11-19T08:13:35.364939639Z"
                            }
                        ],
                        "scheduler_cluster_id": 1
                    }
                ],
                "state": "SUCCESS"
            }
        ],
        "state": "SUCCESS",
        "updated_at": "2024-11-19T08:08:53.157878758Z"
    },
    "scheduler_clusters": [
        {
            "id": 1,
            "created_at": "2024-11-15T08:06:37Z",
            "updated_at": "2024-11-15T08:06:37Z",
            "name": "cluster-1"
        }
    ]
}
```
```

--------------------------------

### Download and Enable Stargz Systemd Service

Source: https://d7y.io/docs/operations/integrations/container-runtime/stargz

Download the stargz-snapshotter systemd service file and enable/start the service.

```bash
wget -O /etc/systemd/system/stargz-snapshotter.service https://raw.githubusercontent.com/containerd/stargz-snapshotter/main/script/config/etc/systemd/system/stargz-snapshotter.service  
systemctl enable --now stargz-snapshotter  
systemctl restart containerd  
```

--------------------------------

### Dragonfly Preheat Job Status Example (Pending)

Source: https://d7y.io/docs/getting-started/quick-start/kubernetes

This JSON shows an example of a preheat job status where the state is 'PENDING'. A 'SUCCESS' state indicates the preheating is complete.

```json
{
  "id": 1,
  "created_at": "2024-04-18T08:51:55Z",
  "updated_at": "2024-04-18T08:51:55Z",
  "task_id": "group_2717f455-ff0a-435f-a3a7-672828d15a2a",
  "type": "preheat",
  "state": "PENDING",
  "args": {
    "filteredQueryParams": "Expires&Signature",
    "headers": null,
    "password": "",
    "pieceLength": 4194304,
    "platform": "",
    "tag": "",
    "type": "image",
    "url": "https://index.docker.io/v2/library/alpine/manifests/3.19",
    "username": ""
  },
  "scheduler_clusters": [
    {
      "id": 1,
      "created_at": "2024-04-18T08:29:15Z",
      "updated_at": "2024-04-18T08:29:15Z",
      "name": "cluster-1"
    }
  ]
}
```

--------------------------------

### Run Scheduler

Source: https://d7y.io/docs/development-guide/configure-development-environment

Execute the Scheduler component using 'go run'. Provide the configuration file path and enable console output.

```bash
# Setup Scheduler.  

go run cmd/scheduler/main.go --config /etc/dragonfly/scheduler.yaml --console  
```

--------------------------------

### Request Trait Definition

Source: https://d7y.io/docs/advanced-guides/request-sdk

Defines the interface for sending GET requests via the Dragonfly SDK. Use `get` for streaming responses and `get_into` to write directly into a buffer.

```rust
/// Defines the interface for sending requests via the Dragonfly.  
///  
/// This trait enables interaction with remote servers through the Dragonfly, providing methods  
/// for performing GET requests with flexible response handling. It is designed for clients that  
/// need to communicate with Dragonfly seed client efficiently, supporting both streaming and buffered  
/// response processing. The trait shields the complex request logic between the client and the  
/// Dragonfly seed client's proxy, abstracting the underlying communication details to simplify  
/// client implementation and usage.  
#[tonic::async_trait]  
pub trait Request {  
    /// Sends an GET request to a remote server via the Dragonfly and returns a response  
    /// with a streaming body.  
    ///  
    /// This method is designed for scenarios where the response body is expected to be processed as a  
    /// stream, allowing efficient handling of large or continuous data. The response includes metadata  
    /// such as status codes and headers, along with a streaming `Body` for accessing the response content.  
    async fn get(&self, request: GetRequest) -> Result<GetResponse<Body>>;  
  
    /// Sends an GET request to a remote server via the Dragonfly and writes the response  
    /// body directly into the provided buffer.  
    ///  
    /// This method is optimized for scenarios where the response body needs to be stored directly in  
    /// memory, avoiding the overhead of streaming for smaller or fixed-size responses. The provided  
    /// `BytesMut` buffer is used to store the response content, and the response metadata (e.g., status  
    /// and headers) is returned separately.  
    async fn get_into(&self, request: GetRequest, buf: &mut BytesMut) -> Result<GetResponse>;  
}

```

--------------------------------

### Install Dragonfly Cluster A with Helm

Source: https://d7y.io/docs/getting-started/quick-start/multi-cluster-kubernetes

Installs Dragonfly cluster A using Helm, applying the specified configuration file. This command ensures the cluster is created and waits for deployment to complete.

```bash
$ helm repo add dragonfly https://dragonflyoss.github.io/helm-charts/  
$ helm install --wait --create-namespace --namespace cluster-a dragonfly dragonfly/dragonfly -f charts-config-cluster-a.yaml  
```

--------------------------------

### Prepare Images for Inference

Source: https://d7y.io/docs/operations/integrations/torchserve

Download sample images (kitten_small.jpg and dogs-before.jpg) to be used for testing the inference API.

```bash
# Prepare pictures that require reasoning.  
curl -O  https://raw.githubusercontent.com/pytorch/serve/master/docs/images/kitten_small.jpg  
curl -O  https://raw.githubusercontent.com/pytorch/serve/master/docs/images/dogs-before.jpg  
```

--------------------------------

### Install Dragonfly Helm Chart

Source: https://d7y.io/docs/operations/integrations/triton-server

Installs the Dragonfly Helm chart into a Kubernetes cluster using a specified configuration file. This command ensures the cluster is created and waits for deployment to complete.

```bash
$ helm repo add dragonfly https://dragonflyoss.github.io/helm-charts/
$ helm install --wait --create-namespace --namespace dragonfly-system dragonfly dragonfly/dragonfly -f charts-config.yaml
LAST DEPLOYED: Mon June 27 19:56:34 2024
NAMESPACE: dragonfly-system
STATUS: deployed
REVISION: 1
TEST SUITE: None
NOTES:
1. Get the scheduler address by running these commands:
  export SCHEDULER_POD_NAME=$(kubectl get pods --namespace dragonfly-system -l "app=dragonfly,release=dragonfly,component=scheduler" -o jsonpath={.items[0].metadata.name})
  export SCHEDULER_CONTAINER_PORT=$(kubectl get pod --namespace dragonfly-system $SCHEDULER_POD_NAME -o jsonpath="{.spec.containers[0].ports[0].containerPort}")
  kubectl --namespace dragonfly-system port-forward $SCHEDULER_POD_NAME 8002:$SCHEDULER_CONTAINER_PORT
  echo "Visit http://127.0.0.1:8002 to use your scheduler"
  
2. Get the dfdaemon port by running these commands:
  export DFDAEMON_POD_NAME=$(kubectl get pods --namespace dragonfly-system -l "app=dragonfly,release=dragonfly,component=dfdaemon" -o jsonpath={.items[0].metadata.name})
  export DFDAEMON_CONTAINER_PORT=$(kubectl get pod --namespace dragonfly-system $DFDAEMON_POD_NAME -o jsonpath="{.spec.containers[0].ports[0].containerPort}")
  You can use $DFDAEMON_CONTAINER_PORT as a proxy port in Node.  
  
3. Configure runtime to use dragonfly:
  https://d7y.io/docs/getting-started/quick-start/kubernetes/  
  
  
4. Get Jaeger query URL by running these commands:
  export JAEGER_QUERY_PORT=$(kubectl --namespace dragonfly-system get services dragonfly-jaeger-query -o jsonpath="{.spec.ports[0].port}")
  kubectl --namespace dragonfly-system port-forward service/dragonfly-jaeger-query 16686:$JAEGER_QUERY_PORT
  echo "Visit http://127.0.0.1:16686/search?limit=20&lookback=1h&maxDuration&minDuration&service=dragonfly to query download events"

```

--------------------------------

### Download Directory with OSS

Source: https://d7y.io/docs/reference/commands/client/dfget

Recursively download an entire directory from an Alibaba Cloud Object Storage Service (OSS) bucket. Authentication credentials and the OSS endpoint are required.

```bash
# Download a directory.  
dfget oss://<bucket>/<path> -O /tmp/path/ --recursive --storage-access-key-id=<access_key_id> --storage-access-key-secret=<access_key_secret> --storage-endpoint=<endpoint>  
```

--------------------------------

### Run Manager

Source: https://d7y.io/docs/development-guide/configure-development-environment

Execute the Manager component using 'go run'. Specify the configuration file path and enable console output.

```bash
# Setup Manager.  

go run cmd/manager/main.go --config /etc/dragonfly/manager.yaml --console  
```

--------------------------------

### Apply Dragonfly Service Configuration

Source: https://d7y.io/docs/operations/integrations/torchserve

Apply the `dfstore.yaml` configuration to create the Kubernetes Service, making the Dragonfly Peer's HTTP proxy accessible.

```bash
kubectl --namespace dragonfly-system apply -f dfstore.yaml
```

--------------------------------

### Download Directory with OBS

Source: https://d7y.io/docs/reference/commands/client/dfget

Recursively download an entire directory from Huawei Cloud Object Storage Service (OBS). Authentication credentials and the OBS endpoint are required.

```bash
# Download a directory.  
dfget obs://<bucket>/<path> -O /tmp/path/ --recursive --storage-access-key-id=<access_key_id> --storage-access-key-secret=<access_key_secret> --storage-endpoint=<endpoint>  
```

--------------------------------

### Download Entire Hugging Face Repository

Source: https://d7y.io/docs/reference/commands/client/dfget

Recursively download an entire repository from Hugging Face. You can specify a revision, defaulting to 'main'.

```bash
# Download an entire repository.  
dfget hf://<owner>/<repo> -O /tmp/repo/ --recursive  

# Download an entire repository from Hugging Face Hub with specified revision. If the revision is not specified, the default value is `main`.  
$ dfget hf://<owner>/<repo> --hf-revision main -O /tmp/repo/ -r  
```

--------------------------------

### Download Verification Model

Source: https://d7y.io/docs/operations/integrations/torchserve

Download a sample model (squeezenet1_1_scripted.mar) from the TorchServe Model Zoo for verification purposes.

```bash
wget https://torchserve.pytorch.org/mar_files/squeezenet1_1_scripted.mar
```

--------------------------------

### Wait for cert-manager to be ready

Source: https://d7y.io/docs/getting-started/installation/helm-charts

Ensure cert-manager is fully available before proceeding with Dragonfly installation.

```bash
kubectl wait --for=condition=Available deployment --all -n cert-manager --timeout=300s
```

--------------------------------

### Configure Peer HTTP Proxy with Basic Authentication

Source: https://d7y.io/docs/operations/best-practices/security

Configure `dfdaemon.yaml` to use Basic Authentication for the peer's HTTP proxy. Note that Basic Auth is not the most secure method and should be used with caution.

```yaml
proxy:  

  # basic_auth is the basic auth configuration for HTTP proxy in dfdaemon. If basic_auth is not  

  # empty, the proxy will use the basic auth to authenticate the client by Authorization  

  # header. The value of the Authorization header is "Basic base64(username:password)", refer  

  # to https://en.wikipedia.org/wiki/Basic_access_authentication.  

  basicAuth:  

    # username is the username for basic auth.  

    username: 'admin'  

    # password is the password for basic auth.  

    password: 'dragonfly'  

```

--------------------------------

### TorchServe Worker Update Response

Source: https://d7y.io/docs/operations/integrations/torchserve

This is the expected response when updating model workers, indicating that the processing has started.

```json
{
  "status": "Processing worker updates..."
}
```

--------------------------------

### Preheat Task Status Example

Source: https://d7y.io/docs/advanced-guides/open-api/preheat

This JSON object represents the status of a preheating task, indicating success.

```json
{
  "id": 1,
  "created_at": "2024-12-11T08:30:12Z",
  "updated_at": "2024-12-11T08:30:51Z",
  "task_id": "group_4dd1da54-96ca-48ff-8f20-4fae665f677f",
  "bio": "",
  "type": "preheat",
  "state": "SUCCESS",
  "args": {
    "platform": "",
    "scope": "single_seed_peer",
    "type": "file",
    "url": "https://example.com"
  },
  "scheduler_clusters": [
    {
      "id": 1,
      "created_at": "2024-12-11T07:57:44Z",
      "updated_at": "2024-12-11T07:57:44Z",
      "name": "cluster-1"
    }
  ]
}
```

--------------------------------

### Create Kind Multi-Node Cluster

Source: https://d7y.io/docs/getting-started/quick-start/kubernetes

Use the 'kind create cluster' command with the specified configuration file to set up a local Kubernetes cluster for testing.

```bash
kind create cluster --config kind-config.yaml
```

--------------------------------

### Verify Seed Peer Ports

Source: https://d7y.io/docs/getting-started/installation/binaries

Verify if the Seed Peer is started and if ports 4000, 4001, and 4002 are available.

```bash
telnet 127.0.0.1 4000  
telnet 127.0.0.1 4001  
telnet 127.0.0.1 4002  

```

--------------------------------

### Create Plugins Path Directory

Source: https://d7y.io/docs/operations/integrations/torchserve

Create a directory to store the binaries of the Dragonfly Endpoint plugin.

```bash
mkdir plugins-path  
```

--------------------------------

### Download a Single File with Authentication

Source: https://d7y.io/docs/operations/integrations/hugging-face

For private repositories or to increase rate limits, provide your Hugging Face token using the --hf-token flag. Ensure the token is kept secure.

```bash
dfget hf://meta-llama/Llama-2-7b/config.json -O /tmp/config.json --hf-token=<token>
```

--------------------------------

### Scheduler CLI Usage

Source: https://d7y.io/docs/reference/commands/scheduler

Basic usage patterns for the scheduler command-line interface. Use this to start the scheduler or access its subcommands.

```bash
scheduler [flags]

scheduler [command]
```

--------------------------------

### Extract Dragonfly Binaries

Source: https://d7y.io/docs/getting-started/installation/binaries

Extract the downloaded Dragonfly binaries to the specified installation directory. Replace '/path/to/dragonfly' with your target path.

```bash
# Replace `/path/to/dragonfly` with the installation directory.  

tar -zxf dragonfly_linux_amd64.tar.gz -C /path/to/dragonfly  
```

--------------------------------

### Download Sample Images for Inference

Source: https://d7y.io/docs/operations/integrations/torchserve

Download sample images to be used for testing the inference API.

```bash
# Prepare pictures that require reasoning.  
curl -O  https://raw.githubusercontent.com/pytorch/serve/master/docs/images/kitten_small.jpg  
curl -O  https://raw.githubusercontent.com/pytorch/serve/master/docs/images/dogs-before.jpg
```

--------------------------------

### Call Inference API

Source: https://d7y.io/docs/operations/integrations/torchserve

Send a request to the TorchServe inference API with the prepared images to get predictions from the 'squeezenet1_1' model.

```bash
# Call inference API.  
curl http://localhost:8080/predictions/squeezenet1_1 -T kitten_small.jpg -T dogs-before.jpg  
```

--------------------------------

### Download Repository Snapshot with Dragonfly

Source: https://d7y.io/docs/operations/integrations/hugging-face

Use this script to download a snapshot of a repository using Dragonfly. Ensure you have Python 3 installed.

```bash
$ python3 snapshot_download_dragonfly.py
```

--------------------------------

### Create and Set Permissions for Model Store

Source: https://d7y.io/docs/operations/integrations/torchserve

Create a directory to store models and set its permissions to 777.

```bash
mkdir model-store  
```

```bash
chmod 777 model-store  
```

--------------------------------

### Call TorchServe Inference API

Source: https://d7y.io/docs/operations/integrations/torchserve

Send requests to the TorchServe inference API with the prepared images. This example uses the squeezenet1_1 model.

```bash
# Call inference API.  
curl http://localhost:8080/predictions/squeezenet1_1 -T kitten_small.jpg -T dogs-before.jpg
```

--------------------------------

### Download File with OSS

Source: https://d7y.io/docs/reference/commands/client/dfget

Download a single file from an Alibaba Cloud Object Storage Service (OSS) bucket. Authentication credentials and the OSS endpoint are required.

```bash
# Download a file.  
dfget oss://<bucket>/<path> -O /tmp/file.txt --storage-access-key-id=<access_key_id> --storage-access-key-secret=<access_key_secret> --storage-endpoint=<endpoint>  
```