mirror of
https://github.com/moby/moby.git
synced 2022-11-09 12:21:53 -05:00
45407cf00a
Docker-DCO-1.1-Signed-off-by: Brian Goff <cpuguy83@gmail.com> (github: cpuguy83)
305 lines
7.8 KiB
Go
305 lines
7.8 KiB
Go
package daemon
|
|
|
|
import (
|
|
"fmt"
|
|
"io/ioutil"
|
|
"os"
|
|
"path/filepath"
|
|
"sort"
|
|
"strings"
|
|
"syscall"
|
|
|
|
"github.com/docker/docker/archive"
|
|
"github.com/docker/docker/daemon/execdriver"
|
|
"github.com/docker/docker/pkg/log"
|
|
"github.com/docker/docker/pkg/symlink"
|
|
"github.com/docker/docker/volumes"
|
|
)
|
|
|
|
type Mount struct {
|
|
MountToPath string
|
|
container *Container
|
|
volume *volumes.Volume
|
|
Writable bool
|
|
}
|
|
|
|
func (container *Container) prepareVolumes() error {
|
|
if container.Volumes == nil || len(container.Volumes) == 0 {
|
|
container.Volumes = make(map[string]string)
|
|
container.VolumesRW = make(map[string]bool)
|
|
if err := container.applyVolumesFrom(); err != nil {
|
|
return err
|
|
}
|
|
}
|
|
|
|
return container.createVolumes()
|
|
}
|
|
|
|
// sortedVolumeMounts returns the list of container volume mount points sorted in lexicographic order
|
|
func (container *Container) sortedVolumeMounts() []string {
|
|
var mountPaths []string
|
|
for path := range container.Volumes {
|
|
mountPaths = append(mountPaths, path)
|
|
}
|
|
|
|
sort.Strings(mountPaths)
|
|
return mountPaths
|
|
}
|
|
|
|
func (container *Container) createVolumes() error {
|
|
mounts, err := container.parseVolumeMountConfig()
|
|
if err != nil {
|
|
return err
|
|
}
|
|
|
|
for _, mnt := range mounts {
|
|
if err := mnt.initialize(); err != nil {
|
|
return err
|
|
}
|
|
}
|
|
|
|
return nil
|
|
}
|
|
|
|
func (m *Mount) initialize() error {
|
|
// No need to initialize anything since it's already been initialized
|
|
if _, exists := m.container.Volumes[m.MountToPath]; exists {
|
|
return nil
|
|
}
|
|
|
|
// This is the full path to container fs + mntToPath
|
|
containerMntPath, err := symlink.FollowSymlinkInScope(filepath.Join(m.container.basefs, m.MountToPath), m.container.basefs)
|
|
if err != nil {
|
|
return err
|
|
}
|
|
m.container.VolumesRW[m.MountToPath] = m.Writable
|
|
m.container.Volumes[m.MountToPath] = m.volume.Path
|
|
m.volume.AddContainer(m.container.ID)
|
|
if m.Writable && !m.volume.IsBindMount {
|
|
// Copy whatever is in the container at the mntToPath to the volume
|
|
copyExistingContents(containerMntPath, m.volume.Path)
|
|
}
|
|
|
|
return nil
|
|
}
|
|
|
|
func (container *Container) VolumePaths() map[string]struct{} {
|
|
var paths = make(map[string]struct{})
|
|
for _, path := range container.Volumes {
|
|
paths[path] = struct{}{}
|
|
}
|
|
return paths
|
|
}
|
|
|
|
func (container *Container) derefVolumes() {
|
|
for path := range container.VolumePaths() {
|
|
vol := container.daemon.volumes.Get(path)
|
|
if vol == nil {
|
|
log.Debugf("Volume %s was not found and could not be dereferenced", path)
|
|
continue
|
|
}
|
|
vol.RemoveContainer(container.ID)
|
|
}
|
|
}
|
|
|
|
func (container *Container) parseVolumeMountConfig() (map[string]*Mount, error) {
|
|
var mounts = make(map[string]*Mount)
|
|
// Get all the bind mounts
|
|
for _, spec := range container.hostConfig.Binds {
|
|
path, mountToPath, writable, err := parseBindMountSpec(spec)
|
|
if err != nil {
|
|
return nil, err
|
|
}
|
|
// Check if a volume already exists for this and use it
|
|
vol := container.daemon.volumes.Get(path)
|
|
if vol == nil {
|
|
vol, err = container.daemon.volumes.NewVolume(path, writable)
|
|
if err != nil {
|
|
return nil, err
|
|
}
|
|
}
|
|
mounts[mountToPath] = &Mount{container: container, volume: vol, MountToPath: mountToPath, Writable: writable}
|
|
}
|
|
|
|
// Get the rest of the volumes
|
|
for path := range container.Config.Volumes {
|
|
// Check if this is already added as a bind-mount
|
|
if _, exists := mounts[path]; exists {
|
|
continue
|
|
}
|
|
|
|
vol, err := container.daemon.volumes.NewVolume("", true)
|
|
if err != nil {
|
|
return nil, err
|
|
}
|
|
mounts[path] = &Mount{container: container, MountToPath: path, volume: vol, Writable: true}
|
|
}
|
|
|
|
return mounts, nil
|
|
}
|
|
|
|
func parseBindMountSpec(spec string) (string, string, bool, error) {
|
|
var (
|
|
path, mountToPath string
|
|
writable bool
|
|
arr = strings.Split(spec, ":")
|
|
)
|
|
|
|
switch len(arr) {
|
|
case 2:
|
|
path = arr[0]
|
|
mountToPath = arr[1]
|
|
writable = true
|
|
case 3:
|
|
path = arr[0]
|
|
mountToPath = arr[1]
|
|
writable = validMountMode(arr[2]) && arr[2] == "rw"
|
|
default:
|
|
return "", "", false, fmt.Errorf("Invalid volume specification: %s", spec)
|
|
}
|
|
|
|
if !filepath.IsAbs(path) {
|
|
return "", "", false, fmt.Errorf("cannot bind mount volume: %s volume paths must be absolute.", path)
|
|
}
|
|
|
|
return path, mountToPath, writable, nil
|
|
}
|
|
|
|
func (container *Container) applyVolumesFrom() error {
|
|
volumesFrom := container.hostConfig.VolumesFrom
|
|
|
|
for _, spec := range volumesFrom {
|
|
mounts, err := parseVolumesFromSpec(container.daemon, spec)
|
|
if err != nil {
|
|
return err
|
|
}
|
|
|
|
for _, mnt := range mounts {
|
|
mnt.container = container
|
|
if err = mnt.initialize(); err != nil {
|
|
return err
|
|
}
|
|
}
|
|
}
|
|
return nil
|
|
}
|
|
|
|
func validMountMode(mode string) bool {
|
|
validModes := map[string]bool{
|
|
"rw": true,
|
|
"ro": true,
|
|
}
|
|
|
|
return validModes[mode]
|
|
}
|
|
|
|
func (container *Container) setupMounts() error {
|
|
mounts := []execdriver.Mount{
|
|
{Source: container.ResolvConfPath, Destination: "/etc/resolv.conf", Writable: true, Private: true},
|
|
}
|
|
|
|
if container.HostnamePath != "" {
|
|
mounts = append(mounts, execdriver.Mount{Source: container.HostnamePath, Destination: "/etc/hostname", Writable: true, Private: true})
|
|
}
|
|
|
|
if container.HostsPath != "" {
|
|
mounts = append(mounts, execdriver.Mount{Source: container.HostsPath, Destination: "/etc/hosts", Writable: true, Private: true})
|
|
}
|
|
|
|
// Mount user specified volumes
|
|
// Note, these are not private because you may want propagation of (un)mounts from host
|
|
// volumes. For instance if you use -v /usr:/usr and the host later mounts /usr/share you
|
|
// want this new mount in the container
|
|
// These mounts must be ordered based on the length of the path that it is being mounted to (lexicographic)
|
|
for _, path := range container.sortedVolumeMounts() {
|
|
mounts = append(mounts, execdriver.Mount{
|
|
Source: container.Volumes[path],
|
|
Destination: path,
|
|
Writable: container.VolumesRW[path],
|
|
})
|
|
}
|
|
|
|
container.command.Mounts = mounts
|
|
return nil
|
|
}
|
|
|
|
func parseVolumesFromSpec(daemon *Daemon, spec string) (map[string]*Mount, error) {
|
|
specParts := strings.SplitN(spec, ":", 2)
|
|
if len(specParts) == 0 {
|
|
return nil, fmt.Errorf("Malformed volumes-from specification: %s", spec)
|
|
}
|
|
|
|
c := daemon.Get(specParts[0])
|
|
if c == nil {
|
|
return nil, fmt.Errorf("Container %s not found. Impossible to mount its volumes", specParts[0])
|
|
}
|
|
|
|
mounts := c.VolumeMounts()
|
|
|
|
if len(specParts) == 2 {
|
|
mode := specParts[1]
|
|
if validMountMode(mode) {
|
|
return nil, fmt.Errorf("Invalid mode for volumes-from: %s", mode)
|
|
}
|
|
|
|
// Set the mode for the inheritted volume
|
|
for _, mnt := range mounts {
|
|
// Ensure that if the inherited volume is not writable, that we don't make
|
|
// it writable here
|
|
mnt.Writable = mnt.Writable && (mode == "rw")
|
|
}
|
|
}
|
|
|
|
return mounts, nil
|
|
}
|
|
|
|
func (container *Container) VolumeMounts() map[string]*Mount {
|
|
mounts := make(map[string]*Mount)
|
|
|
|
for mountToPath, path := range container.Volumes {
|
|
if v := container.daemon.volumes.Get(path); v != nil {
|
|
mounts[mountToPath] = &Mount{volume: v, container: container, MountToPath: mountToPath, Writable: container.VolumesRW[mountToPath]}
|
|
}
|
|
}
|
|
|
|
return mounts
|
|
}
|
|
|
|
func copyExistingContents(source, destination string) error {
|
|
volList, err := ioutil.ReadDir(source)
|
|
if err != nil {
|
|
return err
|
|
}
|
|
|
|
if len(volList) > 0 {
|
|
srcList, err := ioutil.ReadDir(destination)
|
|
if err != nil {
|
|
return err
|
|
}
|
|
|
|
if len(srcList) == 0 {
|
|
// If the source volume is empty copy files from the root into the volume
|
|
if err := archive.CopyWithTar(source, destination); err != nil {
|
|
return err
|
|
}
|
|
}
|
|
}
|
|
|
|
return copyOwnership(source, destination)
|
|
}
|
|
|
|
// copyOwnership copies the permissions and uid:gid of the source file
|
|
// into the destination file
|
|
func copyOwnership(source, destination string) error {
|
|
var stat syscall.Stat_t
|
|
|
|
if err := syscall.Stat(source, &stat); err != nil {
|
|
return err
|
|
}
|
|
|
|
if err := os.Chown(destination, int(stat.Uid), int(stat.Gid)); err != nil {
|
|
return err
|
|
}
|
|
|
|
return os.Chmod(destination, os.FileMode(stat.Mode))
|
|
}
|