Skip to content

Latest commit

 

History

History
506 lines (426 loc) · 15.5 KB

nydusd.md

File metadata and controls

506 lines (426 loc) · 15.5 KB

Nydusd

nydusd running as daemon to expose a FUSE mountpoint or a Virtio-FS mountpoint inside guest for containers to access.

Get binary from release page

Get nydusd binary from release page.

Run Nydusd Daemon

# Prepare nydusd configuration
sudo tee /etc/nydus/nydusd-config.localfs.json > /dev/null << EOF
{
  "device": {
    "backend": {
      "type": "localfs",
      "config": {
        "dir": "/var/lib/nydus/blobs"
      }
    },
    "cache": {
      "type": "blobcache",
      "config": {
        "work_dir": "/var/lib/nydus/cache"
      }
    }
  },
  "mode": "direct",
  "digest_validate": false,
  "iostats_files": false,
  "enable_xattr": true
}

EOF

Run With FUSE

If no /path/to/bootstrap is available, please refer to nydus-image.md for more details.

sudo mkdir -p /var/lib/nydus/blobs/
sudo mkdir -p /var/lib/nydus/cache/
sudo nydusd \
  --config /etc/nydus/nydusd-config.localfs.json \
  --mountpoint /path/to/mnt \
  --bootstrap /path/to/bootstrap \
  --log-level info

For registry backend, we can set authorization with environment variable IMAGE_PULL_AUTH to avoid loading auth from nydusd configuration file.

Run With Virtio-FS

If no /path/to/bootstrap is available, please refer to nydus-image.md for more details.

Virtio-fs is supported by both QEMU and Cloud-hypervisor. To run nydusd with virtio-fs support, first start it with --sock option to expose a virtio-fs socket endpoint.

sudo nydusd \
  --config /etc/nydus/nydusd-config.localfs.json \
  --sock /path/to/vhost-user-fs.sock \
  --bootstrap /path/to/bootstrap \
  --log-level info

Then start a qemu process with a vhost-user-fs-pci device, run something like:

./qemu-system-x86_64 -M pc -cpu host --enable-kvm -smp 2 \
        -m 2G,maxmem=16G -object memory-backend-file,id=mem,size=2G,mem-path=/dev/shm,share=on -numa node,memdev=mem \
        -chardev socket,id=char0,path=/path/to/vhost-user-fs.sock \
        -device vhost-user-fs-pci,chardev=char0,tag=nydus,queue-size=1024,indirect_desc=false,event_idx=false \
        -serial mon:stdio -vga none -nographic -curses -kernel ./kernel \
        -append 'console=ttyS0 root=/dev/vda1 virtio_fs.dyndbg="+pfl" fuse.dyndbg="+pfl"' \
        -device virtio-net-pci,netdev=net0,mac=AE:AD:BE:EF:6C:FB -netdev type=user,id=net0 \
        -qmp unix:/path/to/qmp.sock,server,nowait \
        -drive if=virtio,file=./bionic-server-cloudimg-amd64.img

Then we can mount nydus virtio-fs inside the guest with:

mount -t virtio_fs none /mnt -o tag=nydus,default_permissions,allow_other,rootmode=040000,user_id=0,group_id=0,nodev

Or simply below if you are running newer guest kernel:

mount -t virtiofs nydus /mnt

We are working on enabling cloud-hypervisor support for nydus.

Nydus Configuration

Common Fields In Config

{
  "device": {
    "backend": {
      // localfs | oss | registry
      "type": "localfs",
      "config": {
        // Drop the read request once http request timeout, in seconds
        "timeout": 5,
        // Drop the read request once http connection timeout, in seconds
        "connect_timeout": 5,
        // Retry count when read request failed
        "retry_limit": 0,
      }
    },
    "cache": {
      // Blobcache: enable local fs cache
      // Dummycache: disable cache, access remote storage backend directly
      "type": "blobcache",
      // Enable cache compression
      "compressed": true,
      "config": {
        // Directory of cache files, only for blobcache
        "work_dir": "/cache"
      }
    }
  },
  // direct | cached
  "mode": "direct",
  // Validate inode tree digest and chunk digest on demand
  "digest_validate": false,
  // Enable file IO metric
  "iostats_files": true,
  // Enable support of fs extended attributes
  "enable_xattr": false,
  // Amplified user IO request batch size to read data from remote storage backend / local cache
  // in unit of Bytes, valid values: 0-268435456, default: 1048576
  "amplify_io": 1048576,
  "fs_prefetch": {
    // Enable blob prefetch
    "enable": false,
    // Prefetch thread count
    "threads_count": 10,
    // Maximal read size per prefetch request, e.g. 128kb
    "merging_size": 131072,
    // Limit prefetch bandwidth to 1MB/S, it aims at reducing congestion with normal user io
    "bandwidth_rate": 1048576
  }
}

Use Different Storage Backends

Using different storage backend means that the nydus image metadata (bootstrap) layer is stored in the image registry, but the data layer will be stored on the external storage. Therefore, the option --target for nydusify convert is still required, the registry image reference is needed to store the metadata layer.

Localfs Backend
{
  "device": {
    "backend": {
      "type": "localfs",
      "config": {
        // The directory included all blob files declared in bootstrap
        "dir": "/path/to/blobs/",
        // Record read access log, prefetch data on next time
        "readahead": true,
        // Duration of recording access log
        "readahead_sec": 10
      }
    },
    ...
  },
  ...
}
Localdisk Backend (Experimental)

Using this backend enables Nydus to support reading blobs from block devices. This feature will be useful in Confidential Computing or Hybrid Image scenarios.

The localdisk backend adds support for storing images in disks. In this scenario, each layer of the blob is stored in partitions, and multiple partitions are addressed in the local raw disk via the GUID partition table (GPT), which means that this disk stores the entire image.

Currently, generating a localdisk image through nydusify is not supported for the time being. You need to use the nydus-localdisk tool to complete this step. Document located at: https://github.com/adamqqqplay/nydus-localdisk/blob/master/README.md

{
  "device": {
    "backend": {
      "type": "localdisk",
      "config": {
        // Mounted block device path or original localdisk image file path.
        "device_path": "/dev/loop1"
        //"device_path": "/home/user/ubuntu.img"
      }
    },
    ...
  },
  ...
}
OSS Backend
{
  "device": {
    "backend": {
      "type": "oss",
      "config": {
        ...
        "endpoint": "region.aliyuncs.com",
        "scheme": "https",
        "access_key_id": "",
        "access_key_secret": "",
        "bucket_name": "",
        "object_prefix": "nydus/"
      }
    },
    ...
  },
  ...
}
S3 Backend
{
  "device": {
    "backend": {
      "type": "s3",
      "config": {
        ...
        "endpoint": "s3.amazonaws.com",
        "scheme": "https",
        "access_key_id": "",
        "access_key_secret": "",
        "bucket_name": "",
        "region": "",
        "object_prefix": "nydus/"
      }
    },
    ...
  },
  ...
}
Registry Backend
{
  "device": {
    "backend": {
      "type": "registry",
      "config": {
        ...
        // Registry url scheme, leave empty to automatically detect, otherwise specify to https or http.
        "scheme": "",
        // Registry hostname with format `$host:$port`
        "host": "my-registry:5000",
        // Skip SSL certificate validation for HTTPS scheme
        "skip_verify": false,
        // Use format `$namespace/$repo` (no image tag)
        "repo": "test/repo",
        // Username and password for auth
        // base64(username:password), optional
        "auth": "<base64_encoded_auth>",
        // Bearer token for auth, optional
        "registry_token": "<bearer_token>"
        // Redirected blob download host, optional
        "blob_redirected_host": "<blob_redirected_host>"
      }
    },
    ...
  },
  ...
}

Note: The value of device.backend.config.auth will be overwrite if running the nydusd with environment variable IMAGE_PULL_AUTH.

HTTP Proxy Backend

The HttpProxy backend can access blobs through a http proxy server which can be local (using unix socket) or remote (using https:// or using http://).

HttpProxy uses two API endpoints to access the blobs:

  • HEAD /path/to/blobs to get the blob size
  • GET /path/to/blobs to read the blob

The http proxy server should respect the Range header to compute the offset and length of the blob.

The example config files for the HttpProxy backend may be:

// for remote usage
{
  "device": {
      "backend": {
      "type": "http-proxy",
      "config": {
        "addr": "http://127.0.0.1:9977",
        "path": "/namespace/<repo>/blobs"
      }
    }
  }
}

or

// for local usage
{
  "device": {
      "backend": {
      "type": "http-proxy",
      "config": {
        "addr": "/path/to/unix.sock",
      }
    }
  }
}

The HttpProxy backend also supports the Proxy and Mirrors configurations for remote usage like the Registry backend described above.

Enable Mirrors for Storage Backend (Recommend)

Nydus is deeply integrated with Dragonfly P2P mirror mode, please refer the doc to learn how configuring Nydus to use Dragonfly.

Add device.backend.config.mirrors field to enable mirrors for storage backend. The mirror can be a P2P distribution server or registry. If the request to mirror server failed, it will fall back to the original registry. Currently, the mirror mode is only tested in the registry backend, and in theory, the OSS backend also supports it.

!! The mirrors field conflicts with proxy field.

{
  "device": {
    "backend": {
      "type": "registry",
      "config": {
        "mirrors": [
          {
            // Mirror server URL (including scheme), e.g. Dragonfly dfdaemon server URL
            "host": "http://dragonfly1.io:65001",
            // Headers for mirror server
            "headers": {
              // For Dragonfly dfdaemon server URL, we need to specify "X-Dragonfly-Registry" (including scheme).
              // When Dragonfly does not cache data, nydusd will pull it from "X-Dragonfly-Registry".
              // If not set "X-Dragonfly-Registry", Dragonfly will pull data from proxy.registryMirror.url.
              "X-Dragonfly-Registry": "https://index.docker.io"
            },
            // This URL endpoint is used to check the health of mirror server, and if the mirror is unhealthy, 
            // the request will fallback to the next mirror or the original registry server. 
            // Use $host/v2 as default if left empty.
            "ping_url": "http://127.0.0.1:40901/server/ping",
            // Interval time (s) to check and recover unavailable mirror. Use 5 as default if left empty.
            "health_check_interval": 5,
            // Failure counts before disabling this mirror. Use 5 as default if left empty.
            "failure_limit": 5,
          },
          {
            "host": "http://dragonfly2.io:65001",
            "headers": {
              "X-Dragonfly-Registry": "https://index.docker.io"
            },
          }
        ],
        ...
      }
    },
    ...
  },
  ...
}
Enable P2P Proxy for Storage Backend

Add device.backend.config.proxy field to enable HTTP proxy for storage backend. For example, use P2P distribution service to reduce network workload and latency in large scale container cluster using Dragonfly (enable centralized dfdaemon mode).

{
  "device": {
    "backend": {
      "type": "registry",
      "config": {
        "proxy": {
          // Access remote storage backend via P2P proxy, e.g. Dragonfly dfdaemon server URL
          "url": "http://p2p-proxy:65001",
          // Fallback to remote storage backend if P2P proxy ping failed
          "fallback": true,
          // Endpoint of P2P proxy health checking
          "ping_url": "http://p2p-proxy:40901/server/ping",
          // Interval of P2P proxy health checking, in seconds
          "check_interval": 5
        },
        ...
      }
    },
    ...
  },
  ...
}

Once the configuration is loaded successfully on nydusd starting, we will see the log as shown below:

INFO [storage/src/backend/connection.rs:136] backend config: CommonConfig { proxy: ProxyConfig { url: "http://p2p-proxy:65001", ping_url: "http://p2p-proxy:40901/server/ping", fallback: true, check_interval: 5 }, timeout: 5, connect_timeout: 5, retry_limit: 0 }

Mount writable Overlay FS

Nydusd itself has a native userspace Overlay FS implementation, which can be enabled with several extra configurations.

An example configuration /etc/nydus/nydusd-config.overlay.json is as follows:

{
	"version": 2,
	"backend": {
		"type": "localfs",
		"localfs": "/var/lib/nydus/blobs"
	},
	"cache": {
		"type": "blobcache",
		"filecache": {
			"work_dir": "/var/lib/nydus/cache"
		}
	},
	"rafs": {
		"mode": "direct",
		"enable_xattr": true
	},
	"overlay": {
		"upper_dir": "/path/to/upperdir",
		"work_dir": "/path/to/workdir"
	}
}

An extra field overlay is added to the Nydusd configuration, which specifies the upper_dir and work_dir for the overlay filesystem.

You can start nydusd with the above configuration and such command:

sudo nydusd \
    --config /etc/nydus/nydusd-config.overlay.json \
    --mountpoint /path/to/mnt/ \
    --bootstrap /path/to/bootstrap \
    --log-level info \
    --writable

This will create a FUSE overlay mountpoint at /path/to/mnt/, with one Nydus image as readonly lower layer and the /path/to/upperdir as writable upper layer, so that it can take over whole rootfs of a container, any contents writen from container will be stored in /path/to/upperdir.

Removing --writable flag will make the overlay filesystem readonly if you wish.

Mount Bootstrap Via API

To mount a bootstrap via api, first launch nydusd without a bootstrap:

sudo nydusd \
  --apisock /path/to/api.sock \
  --config /path/to/config.json \
  --mountpoint /path/to/mountpoint

Then use curl to mount a bootstrap to /path/to/mountpoint/sub:

curl --unix-socket api.sock \
     -X POST "http://localhost/api/v1/mount?mountpoint=/sub" \
     -H "Content-Type: application/json" \
     -d '{
        "source":"/path/to/bootstrap",
        "fs_type":"rafs",
        "config":"{\"device\":{\"backend\":{\"type\":\"localfs\",\"config\":{\"dir\":\"blobs\"}},\"cache\":{\"type\":\"blobcache\",\"config\":{\"work_dir\":\"cache\"}}},\"mode\":\"direct\",\"digest_validate\":true}"
	}'

The config field is a JSON format string that can be obtained by cat rafs.config | jq tostring.

Multiple Pseudo Mounts

One single nydusd can have multiple pseudo mounts within a mountpoint.

To achieve that, you can trigger backend fs (e.g., rafs) mount through the HTTP interfaces using curl command.

When starting nydusd without the --bootstrap option, there will be no backend file system in a nydus mountpoint. You can use curl command to mount multiple backend fs at different sub-directories.

Example

Given that your mountpoint is /mnt which can be a directory in local host or inside guest.

When you have two pseudo mounts which are named "pseudo_1" and "pseudo_2" identified in http request body.

pseudo_1 and pseudo_2 correspond to bootstrap respectively.

tree -L 1 mnt
mnt
├── pseudo_1
└── pseudo_2