rbd

func ExportBackup ¶

func ExportBackup(ctx context.Context, client testcontainers.Container, image string, destination io.Writer) error

ExportBackup writes a CLI-generated format-2 archive to destination. The archive includes image metadata and snapshots. No mirror daemon is required. The caller owns the writer and decides where to retain the backup.

client is any container that has the rbd CLI and credentials for its own cluster, such as ceph.Container.ClientContainer or a container attached with WithClient, and image is an image spec such as volumes/disk or volumes/tenant-a/disk. The archive is written to a temporary file under /tmp in the client container and removed afterwards. The backup functions never close the caller's reader or writer, and they check cancellation between reads, so a reader or writer that can block indefinitely must support its own cancellation.

func ExportIncremental ¶

func ExportIncremental(ctx context.Context, client testcontainers.Container, image, fromSnapshot string, destination io.Writer) error

ExportIncremental writes the CLI diff from fromSnapshot to image (which can include @snapshot). Restore requires the matching baseline snapshot. fromSnapshot is a bare snapshot name. With image@snapshot the diff ends at that snapshot, and restoring it creates that snapshot on the target. The client and stream rules are those of ExportBackup.

func InitPool ¶

func InitPool(ctx context.Context, c *ceph.Container, pool string) error

InitPool initializes an existing replicated pool for RBD metadata. Pools registered to other applications are rejected; no force option is used. An existing RBD initialization is safe to repeat. EC pools may supply image data but cannot hold RBD metadata. Create the pool first, then wait for clean PGs before client I/O. Partial initialization never deletes pool data. Success is verified natively, by the pool's application tag and its rbd_trash object, not only by the CLI exit code. Pool names may contain letters, digits, _, . and -, must not start with . or -, and are at most 128 characters.

package main

import (
	"context"
	"log"

	"github.com/JSYoo5B/ceph-testcontainers-go/ceph"
	"github.com/JSYoo5B/ceph-testcontainers-go/rbd"
	"github.com/testcontainers/testcontainers-go"
)

func main() {
	ctx := context.Background()
	cluster, err := ceph.Run(ctx, ceph.DefaultImage,
		ceph.WithClusterConfig(ceph.ClusterConfig{}.WithPools(ceph.PoolConfig{Name: "images", Application: "rbd"})),
	)
	defer func() {
		if err := testcontainers.TerminateContainer(cluster); err != nil {
			log.Printf("failed to terminate container: %s", err)
		}
	}()
	if err != nil {
		panic(err)
	}
	if err := rbd.InitPool(ctx, cluster, "images"); err != nil {
		panic(err)
	}
	namespace, err := rbd.CreateNamespace(ctx, cluster, "images", "tenant-a")
	if err != nil {
		panic(err)
	}
	// Pass namespace.PoolName() and namespace.Name() to a native RBD client.
	// Images, snapshots and trash must be removed before deleting the namespace.
	if err := rbd.RemoveNamespace(ctx, cluster, namespace); err != nil {
		panic(err)
	}
}

Output:

func ListNamespaces ¶

func ListNamespaces(ctx context.Context, c *ceph.Container, pool string) ([]string, error)

ListNamespaces returns the pool's native namespace names in sorted order, including names created outside this fixture. The unnamed default namespace is omitted by Ceph. Listing does not grant ownership for removal.

func RemoveNamespace ¶

func RemoveNamespace(ctx context.Context, c *ceph.Container, ns *Namespace) error

RemoveNamespace removes a confirmed namespace owned by this cluster. Ceph refuses nonempty namespaces, including images in trash. No image purge or force option is used. Successful removal is idempotent; failed CLI removal can be retried with the same descriptor and a new context. Replacing its underlying pool is rejected. Do not delete and recreate the same named namespace through another client while its descriptor is in use: native RBD namespaces have no separate generation identity to distinguish replacements.

func RestoreBackup ¶

func RestoreBackup(ctx context.Context, client testcontainers.Container, image string, archive io.Reader) error

RestoreBackup imports a format-2 archive into a new image on the client's own cluster. Pool creation and archive retention belong to the caller. It first spools the reader into a private temporary file on the test host, so a large archive is not held in memory, then copies it into the container; the client and stream rules are those of ExportBackup.

func RestoreIncremental ¶

func RestoreIncremental(ctx context.Context, client testcontainers.Container, image string, archive io.Reader) error

RestoreIncremental applies a CLI diff to an existing image. Ceph checks that its baseline snapshot exists; this function does not invent a baseline. The check is by snapshot name only: nothing proves that the baseline has the same content.

func WithPools ¶

func WithPools(configs ...ceph.PoolConfig) ceph.Option

WithPools creates replicated pools for RBD images during Run and initializes their RBD metadata. Application defaults to rbd and must not be another application; erasure-coded data pools are created with ceph.ClusterConfig.WithPools and referenced as an image's data pool. Repeated options replace the list, and passing the option replaces the default rbd pool of Run. The other ceph.PoolConfig fields keep their ceph meaning: a zero PGNum selects 8 PGs with the autoscaler off, zero Replicas and MinSize select the cluster defaults (2/1, or 1/1 with one OSD), and the placement fields are validated against the OSDs like those of any other pool.