Chapter 3. Live migration of images

By default, during the live migration of the RBD images with the same storage cluster, the source image is marked read-only. All clients redirect the Input/Output (I/O) to the new target image. Additionally, this mode can preserve the link to the source image’s parent to preserve sparseness, or it can flatten the image during the migration to remove the dependency on the source image’s parent. You can use the live migration process in an import-only mode, where the source image remains unmodified. You can link the target image to an external data source, such as a backup file, HTTP(s) file, or an S3 object. The live migration copy process can safely run in the background while the new target image is being used.

The live migration process consists of three steps:

Prepare Migration: The first step is to create new target image and link the target image to the source image. If the import-only mode is not configured, the source image will also be linked to the target image and marked read-only. Attempts to read uninitialized data extents within the target image will internally redirect the read to the source image, and writes to uninitialized extents within the target image will internally deep copy, the overlapping source image extents to the target image.

Execute Migration: This is a background operation that deep-copies all initialized blocks from the source image to the target. You can run this step when clients are actively using the new target image.

Finish Migration: You can commit or abort the migration, once the background migration process is completed. Committing the migration removes the cross-links between the source and target images, and will remove the source image if not configured in the import-only mode. Aborting the migration remove the cross-links, and will remove the target image.

You can use the native format to describe a native RBD image within a Red Hat Ceph Storage cluster as the source image. The source-spec JSON document is encoded as:

{
    "type": "native",
    "pool_name": "POOL_NAME",
    ["pool_id": "POOL_ID",] (optional, alternative to "POOL_NAME" key)
    ["pool_namespace": "POOL_NAMESPACE",] (optional)
    "image_name": "IMAGE_NAME>",
    ["image_id": "IMAGE_ID",] (optional, useful if image is in trash)
    "snap_name": "SNAP_NAME",
    ["snap_id": "SNAP_ID",] (optional, alternative to "SNAP_NAME" key)
}

Note that the native format does not include the stream object since it utilizes native Ceph operations. For example, to import from the image rbd/ns1/image1@snap1, the source-spec could be encoded as:

{
"type": "native",
"pool_name": "rbd",
"pool_namespace": "ns1",
"image_name": "image1",
"snap_name": "snap1"
}

You can use the qcow format to describe a QEMU copy-on-write (QCOW) block device. Both the QCOW v1 and v2 formats are currently supported with the exception of advanced features such as compression, encryption, backing files, and external data files. You can link the qcow format data to any supported stream source:

{
    "type": "qcow",
    "stream": {
      "type": "file",
  "file_path": "/mnt/image.qcow"
    }
}

You can use the raw format to describe a thick-provisioned, raw block device export that is rbd export –export-format 1 SNAP_SPEC. You can link the raw format data to any supported stream source:

{
    "type": "raw",
    "stream": {
      "type": "file",
      "file_path": "/mnt/image-head.raw"
    },
    "snapshots": [
        {
            "type": "raw",
            "name": "snap1",
            "stream": {
              "type": "file",
       "file_path": "/mnt/image-snap1.raw"
            }
        },
    ] (optional oldest to newest ordering of snapshots)
}

The inclusion of the snapshots array is optional and currently only supports thick-provisioned raw snapshot exports.

{
    <format unique parameters>
    "stream": {
        "type": "file",
        "file_path": "FILE_PATH"
    }
}

For example, to import a raw-format image from a file located at /mnt/image.raw, the source-spec JSON file is:

{
    "type": "raw",
    "stream": {
        "type": "file",
        "file_path": "/mnt/image.raw"
    }
}

{
    <format unique parameters>
    "stream": {
        "type": "http",
        "url": "URL_PATH"
    }
}

For example, to import a raw-format image from a file located at http://download.ceph.com/image.raw, the source-spec JSON file is:

{
    "type": "raw",
    "stream": {
        "type": "http",
        "url": "http://download.ceph.com/image.raw"
    }
}

{
    <format unique parameters>
    "stream": {
        "type": "s3",
        "url": "URL_PATH",
        "access_key": "ACCESS_KEY",
        "secret_key": "SECRET_KEY"
    }
}

For example, to import a raw-format image from a file located at http://s3.ceph.com/bucket/image.raw, its source-spec JSON is encoded as follows:

{
    "type": "raw",
    "stream": {
        "type": "s3",
        "url": "http://s3.ceph.com/bucket/image.raw",
        "access_key": "NX5QOQKC6BH2IDN8HC7A",
        "secret_key": "LnEsqNNqZIpkzauboDcLXLcYaWwLQ3Kop0zAnKIn"
    }
}

You can prepare the default live migration process for RBD images within the same Red Hat Ceph Storage cluster. The rbd migration prepare command accepts all the same layout options as the rbd create command. The rbd create command allows changes to the on-disk layout of the immutable image. If you only want to change the on-disk layout and want to keep the original image name, skip the migration_target argument. All clients using the source image must be stopped before preparing a live migration. The prepare step will fail if it finds any running clients with the image open in read/write mode. You can restart the clients using the new target image once the prepare step is completed.

You can initiate the import-only live migration process by running the rbd migration prepare command with the --import-only and either, --source-spec or --source-spec-path options, passing a JSON document that describes how to access the source image data directly on the command line or from a file.

After you prepare for the live migration, you must copy the image blocks from the source image to the target image.

You can commit the migration, once the live migration has completed deep-copying all the data blocks from the source image to the target image.

[ceph: root@rbd-client /]# rbd trash list --all sourcepool1

You can revert the live migration process. Aborting live migration reverts the prepare and execute steps.

[ceph: root@rbd-client /]# rbd ls sourcepool1
sourceimage1