Documentation
¶
Index ¶
- func NewCreateCmd(globalFlags *flags.GlobalFlags) *cobra.Command
- func NewDeleteCmd(globalFlags *flags.GlobalFlags) *cobra.Command
- func NewListCmd(globalFlags *flags.GlobalFlags) *cobra.Command
- func NewRestoreCmd(globalFlags *flags.GlobalFlags) *cobra.Command
- func NewSnapshotCmd(globalFlags *flags.GlobalFlags) *cobra.Command
- type CreateCmd
- type DeleteCmd
- type ListCmd
- type RestoreCmd
Constants ¶
This section is empty.
Variables ¶
This section is empty.
Functions ¶
func NewCreateCmd ¶
func NewCreateCmd(globalFlags *flags.GlobalFlags) *cobra.Command
func NewDeleteCmd ¶
func NewDeleteCmd(globalFlags *flags.GlobalFlags) *cobra.Command
func NewListCmd ¶
func NewListCmd(globalFlags *flags.GlobalFlags) *cobra.Command
func NewRestoreCmd ¶
func NewRestoreCmd(globalFlags *flags.GlobalFlags) *cobra.Command
func NewSnapshotCmd ¶
func NewSnapshotCmd(globalFlags *flags.GlobalFlags) *cobra.Command
Types ¶
type CreateCmd ¶
type CreateCmd struct {
*flags.GlobalFlags
Registry string
Message string
}
func (*CreateCmd) Run ¶
The manifest is pushed last so a snapshot only becomes visible once every blob it references already exists in the registry.
The workspace is not quiesced (stopped/paused) while its filesystem is committed and its volumes are read: the two are captured back-to-back without a consistency window, so a write straddling both (e.g. a build tool that writes to the container filesystem and a bind-mounted volume in the same operation) can be observed half-applied in the resulting snapshot. This is an accepted limitation, not an oversight: volumes are read directly off host bind mount paths rather than through the container (see pushVolumes), so stopping the container would not by itself close this window, and would also interrupt the workspace's running agent/SSH session for a `snapshot create` that today runs against a live workspace.
type DeleteCmd ¶
type DeleteCmd struct {
*flags.GlobalFlags
}
type ListCmd ¶
type ListCmd struct {
*flags.GlobalFlags
Registry string
}
type RestoreCmd ¶
type RestoreCmd struct {
*flags.GlobalFlags
WorkspaceID string
ProviderName string
}
func (*RestoreCmd) Run ¶
func (cmd *RestoreCmd) Run( ctx context.Context, devsyConfig *config.Config, snapshotRef string, ) error
Run reuses the same creation/start path `devsy up` uses (pkg/workspace.Resolve plus the agent run), with the workspace's source pinned to the snapshot ref and the devcontainer image pinned to the snapshot's committed filesystem layer, so the snapshot volume restore wired into container setup (cmd/internal/agentcontainer/setup.go) runs unchanged.
Source Files