mount

package
v1.8.9-rancher1 Latest Latest
Warning

This package is not in the latest version of its module.

 Go to latest Published: Mar 12, 2018 License: Apache-2.0 Imports: 15 Imported by: 0

Documentation

Overview

Package mount defines an interface to mounting filesystems.

TODO(thockin): This whole pkg is pretty linux-centric. As soon as we have an alternate platform, we will need to abstract further.

Index

Constants

View Source 
const FakeActionMount = "mount"

Values for FakeAction.Action

View Source 
const FakeActionUnmount = "unmount"
View Source 
const (
	MountsInGlobalPDPath = "mounts"
)

Variables

This section is empty.

Functions

func GetDeviceNameFromMount added in v0.15.0 

func GetDeviceNameFromMount(mounter Interface, mountPath string) (string, int, error)

GetDeviceNameFromMount: given a mnt point, find the device from /proc/mounts returns the device name, reference count, and error code

func GetMountRefs added in v0.13.0 

func GetMountRefs(mounter Interface, mountPath string) ([]string, error)

GetMountRefs finds all other references to the device referenced by mountPath; returns a list of paths.

func GetMountRefsByDev added in v1.8.0 

func GetMountRefsByDev(mounter Interface, mountPath string) ([]string, error)

GetMountRefsByDev finds all references to the device provided by mountPath; returns a list of paths.

func IsNotMountPoint added in v1.4.5 

func IsNotMountPoint(mounter Interface, file string) (bool, error)

IsNotMountPoint determines if a directory is a mountpoint. It should return ErrNotExist when the directory does not exist. This method uses the List() of all mountpoints It is more extensive than IsLikelyNotMountPoint and it detects bind mounts in linux

Types

type Exec added in v1.8.0 

type Exec interface {
	// Run executes a command and returns its stdout + stderr combined in one
	// stream.
	Run(cmd string, args ...string) ([]byte, error)
}

Exec executes command where mount utilities are. This can be either the host, container where kubelet runs or even a remote pod with mount utilities. Usual pkg/util/exec interface is not used because kubelet.RunInContainer does not provide stdin/stdout/stderr streams.

func NewOsExec added in v1.8.0 

func NewOsExec() Exec

type FakeAction added in v0.13.1 

type FakeAction struct {
	Action string // "mount" or "unmount"
	Target string // applies to both mount and unmount actions
	Source string // applies only to "mount" actions
	FSType string // applies only to "mount" actions
}

FakeAction objects are logged every time a fake mount or unmount is called.

type FakeExec added in v1.8.0 

type FakeExec struct {
	// contains filtered or unexported fields
}

Fake for testing.

func NewFakeExec added in v1.8.0 

func NewFakeExec(run runHook) *FakeExec

func (*FakeExec) Run added in v1.8.0 

func (f *FakeExec) Run(cmd string, args ...string) ([]byte, error)

type FakeMounter added in v0.13.0 

type FakeMounter struct {
	MountPoints []MountPoint
	Log         []FakeAction
	// contains filtered or unexported fields
}

FakeMounter implements mount.Interface for tests.

func (*FakeMounter) CleanSubPaths added in v1.8.9 

func (f *FakeMounter) CleanSubPaths(podDir string, volumeName string) error

func (*FakeMounter) DeviceOpened added in v1.4.0 

func (f *FakeMounter) DeviceOpened(pathname string) (bool, error)

func (*FakeMounter) GetDeviceNameFromMount added in v1.4.0 

func (f *FakeMounter) GetDeviceNameFromMount(mountPath, pluginDir string) (string, error)

func (*FakeMounter) IsLikelyNotMountPoint added in v1.1.0 

func (f *FakeMounter) IsLikelyNotMountPoint(file string) (bool, error)

func (*FakeMounter) IsMountPointMatch added in v1.7.2 

func (f *FakeMounter) IsMountPointMatch(mp MountPoint, dir string) bool

func (*FakeMounter) IsNotMountPoint added in v1.7.2 

func (f *FakeMounter) IsNotMountPoint(dir string) (bool, error)

func (*FakeMounter) List added in v0.13.0 

func (f *FakeMounter) List() ([]MountPoint, error)

func (*FakeMounter) MakeRShared added in v1.8.0 

func (f *FakeMounter) MakeRShared(path string) error

func (*FakeMounter) Mount added in v0.13.0 

func (f *FakeMounter) Mount(source string, target string, fstype string, options []string) error

func (*FakeMounter) PathIsDevice added in v1.4.0 

func (f *FakeMounter) PathIsDevice(pathname string) (bool, error)

func (*FakeMounter) PrepareSafeSubpath added in v1.8.9 

func (f *FakeMounter) PrepareSafeSubpath(subPath Subpath) (newHostPath string, cleanupAction func(), err error)

func (*FakeMounter) ResetLog added in v0.13.1 

func (f *FakeMounter) ResetLog()

func (*FakeMounter) SafeMakeDir added in v1.8.9 

func (mounter *FakeMounter) SafeMakeDir(pathname string, base string, perm os.FileMode) error

func (*FakeMounter) Unmount added in v0.13.0 

func (f *FakeMounter) Unmount(target string) error

type Interface 

type Interface interface {
	// Mount mounts source to target as fstype with given options.
	Mount(source string, target string, fstype string, options []string) error
	// Unmount unmounts given target.
	Unmount(target string) error
	// List returns a list of all mounted filesystems.  This can be large.
	// On some platforms, reading mounts is not guaranteed consistent (i.e.
	// it could change between chunked reads). This is guaranteed to be
	// consistent.
	List() ([]MountPoint, error)
	// IsMountPointMatch determines if the mountpoint matches the dir
	IsMountPointMatch(mp MountPoint, dir string) bool
	// IsNotMountPoint determines if a directory is a mountpoint.
	// It should return ErrNotExist when the directory does not exist.
	// IsNotMountPoint is more expensive than IsLikelyNotMountPoint.
	// IsNotMountPoint detects bind mounts in linux.
	// IsNotMountPoint enumerates all the mountpoints using List() and
	// the list of mountpoints may be large, then it uses
	// IsMountPointMatch to evaluate whether the directory is a mountpoint
	IsNotMountPoint(file string) (bool, error)
	// IsLikelyNotMountPoint uses heuristics to determine if a directory
	// is a mountpoint.
	// It should return ErrNotExist when the directory does not exist.
	// IsLikelyNotMountPoint does NOT properly detect all mountpoint types
	// most notably linux bind mounts.
	IsLikelyNotMountPoint(file string) (bool, error)
	// DeviceOpened determines if the device is in use elsewhere
	// on the system, i.e. still mounted.
	DeviceOpened(pathname string) (bool, error)
	// PathIsDevice determines if a path is a device.
	PathIsDevice(pathname string) (bool, error)
	// GetDeviceNameFromMount finds the device name by checking the mount path
	// to get the global mount path which matches its plugin directory
	GetDeviceNameFromMount(mountPath, pluginDir string) (string, error)
	// MakeRShared checks that given path is on a mount with 'rshared' mount
	// propagation. If not, it bind-mounts the path as rshared.
	MakeRShared(path string) error
	// SafeMakeDir makes sure that the created directory does not escape given
	// base directory mis-using symlinks. The directory is created in the same
	// mount namespace as where kubelet is running. Note that the function makes
	// sure that it creates the directory somewhere under the base, nothing
	// else. E.g. if the directory already exists, it may exists outside of the
	// base due to symlinks.
	SafeMakeDir(pathname string, base string, perm os.FileMode) error
	// CleanSubPaths removes any bind-mounts created by PrepareSafeSubpath in given
	// pod volume directory.
	CleanSubPaths(podDir string, volumeName string) error
	// PrepareSafeSubpath does everything that's necessary to prepare a subPath
	// that's 1) inside given volumePath and 2) immutable after this call.
	//
	// newHostPath - location of prepared subPath. It should be used instead of
	// hostName when running the container.
	// cleanupAction - action to run when the container is running or it failed to start.
	//
	// CleanupAction must be called immediately after the container with given
	// subpath starts. On the other hand, Interface.CleanSubPaths must be called
	// when the pod finishes.
	PrepareSafeSubpath(subPath Subpath) (newHostPath string, cleanupAction func(), err error)
}

func New 

func New(mounterPath string) Interface

New returns a mount.Interface for the current system. It provides options to override the default mounter behavior. mounterPath allows using an alternative to `/bin/mount` for mounting.

type MountPoint 

type MountPoint struct {
	Device string
	Path   string
	Type   string
	Opts   []string
	Freq   int
	Pass   int
}

This represents a single line in /proc/mounts or /etc/fstab.

type Mounter added in v0.10.0 

type Mounter struct {
	// contains filtered or unexported fields
}

Mounter provides the default implementation of mount.Interface for the linux platform. This implementation assumes that the kubelet is running in the host's root mount namespace.

func (*Mounter) CleanSubPaths added in v1.8.9 

func (mounter *Mounter) CleanSubPaths(podDir string, volumeName string) error

func (*Mounter) DeviceOpened added in v1.4.0 

func (mounter *Mounter) DeviceOpened(pathname string) (bool, error)

DeviceOpened checks if block device in use by calling Open with O_EXCL flag. If pathname is not a device, log and return false with nil error. If open returns errno EBUSY, return true with nil error. If open returns nil, return false with nil error. Otherwise, return false with error

func (*Mounter) GetDeviceNameFromMount added in v1.4.0 

func (mounter *Mounter) GetDeviceNameFromMount(mountPath, pluginDir string) (string, error)

GetDeviceNameFromMount: given a mount point, find the device name from its global mount point

func (*Mounter) IsLikelyNotMountPoint added in v1.1.0 

func (mounter *Mounter) IsLikelyNotMountPoint(file string) (bool, error)

IsLikelyNotMountPoint determines if a directory is not a mountpoint. It is fast but not necessarily ALWAYS correct. If the path is in fact a bind mount from one part of a mount to another it will not be detected. mkdir /tmp/a /tmp/b; mount --bin /tmp/a /tmp/b; IsLikelyNotMountPoint("/tmp/b") will return true. When in fact /tmp/b is a mount point. If this situation if of interest to you, don't use this function...

func (*Mounter) IsMountPointMatch added in v1.7.2 

func (mounter *Mounter) IsMountPointMatch(mp MountPoint, dir string) bool

func (*Mounter) IsNotMountPoint added in v1.7.2 

func (mounter *Mounter) IsNotMountPoint(dir string) (bool, error)

func (*Mounter) List added in v0.10.0 

func (*Mounter) List() ([]MountPoint, error)

List returns a list of all mounted filesystems.

func (*Mounter) MakeRShared added in v1.8.0 

func (mounter *Mounter) MakeRShared(path string) error

func (*Mounter) Mount added in v0.10.0 

func (mounter *Mounter) Mount(source string, target string, fstype string, options []string) error

Mount mounts source to target as fstype with given options. 'source' and 'fstype' must be an emtpy string in case it's not required, e.g. for remount, or for auto filesystem type, where kernel handles fs type for you. The mount 'options' is a list of options, currently come from mount(8), e.g. "ro", "remount", "bind", etc. If no more option is required, call Mount with an empty string list or nil.

func (*Mounter) PathIsDevice added in v1.4.0 

func (mounter *Mounter) PathIsDevice(pathname string) (bool, error)

PathIsDevice uses FileInfo returned from os.Stat to check if path refers to a device.

func (*Mounter) PrepareSafeSubpath added in v1.8.9 

func (mounter *Mounter) PrepareSafeSubpath(subPath Subpath) (newHostPath string, cleanupAction func(), err error)

func (*Mounter) SafeMakeDir added in v1.8.9 

func (mounter *Mounter) SafeMakeDir(pathname string, base string, perm os.FileMode) error

func (*Mounter) Unmount added in v0.10.0 

func (mounter *Mounter) Unmount(target string) error

Unmount unmounts the target.

type NsenterMounter added in v0.17.0 

type NsenterMounter struct {
	// contains filtered or unexported fields
}

NsenterMounter is part of experimental support for running the kubelet in a container. Currently, all docker containers receive their own mount namespaces. NsenterMounter works by executing nsenter to run commands in the host's mount namespace.

NsenterMounter requires:

  1. Docker >= 1.6 due to the dependency on the slave propagation mode of the bind-mount of the kubelet root directory in the container. Docker 1.5 used a private propagation mode for bind-mounts, so mounts performed in the host's mount namespace do not propagate out to the bind-mount in this docker version.
  2. The host's root filesystem must be available at /rootfs
  3. The nsenter binary must be on the Kubelet process' PATH in the container's filesystem.
  4. The Kubelet process must have CAP_SYS_ADMIN (required by nsenter); at the present, this effectively means that the kubelet is running in a privileged container.
  5. The volume path used by the Kubelet must be the same inside and outside the container and be writable by the container (to initialize volume) contents. TODO: remove this requirement.
  6. The host image must have mount, findmnt, and umount binaries in /bin, /usr/sbin, or /usr/bin
  7. The host image should have systemd-run in /bin, /usr/sbin, or /usr/bin

For more information about mount propagation modes, see: 

https://www.kernel.org/doc/Documentation/filesystems/sharedsubtree.txt

func NewNsenterMounter added in v1.0.6 

func NewNsenterMounter() *NsenterMounter

func (*NsenterMounter) CleanSubPaths added in v1.8.9 

func (mounter *NsenterMounter) CleanSubPaths(podDir string, volumeName string) error

func (*NsenterMounter) DeviceOpened added in v1.4.0 

func (n *NsenterMounter) DeviceOpened(pathname string) (bool, error)

DeviceOpened checks if block device in use by calling Open with O_EXCL flag. Returns true if open returns errno EBUSY, and false if errno is nil. Returns an error if errno is any error other than EBUSY. Returns with error if pathname is not a device.

func (*NsenterMounter) GetDeviceNameFromMount added in v1.4.0 

func (n *NsenterMounter) GetDeviceNameFromMount(mountPath, pluginDir string) (string, error)

GetDeviceNameFromMount given a mount point, find the volume id from checking /proc/mounts

func (*NsenterMounter) IsLikelyNotMountPoint added in v1.1.0 

func (n *NsenterMounter) IsLikelyNotMountPoint(file string) (bool, error)

IsLikelyNotMountPoint determines whether a path is a mountpoint by calling findmnt in the host's root mount namespace.

func (*NsenterMounter) IsMountPointMatch added in v1.7.2 

func (*NsenterMounter) IsMountPointMatch(mp MountPoint, dir string) bool

func (*NsenterMounter) IsNotMountPoint added in v1.7.2 

func (m *NsenterMounter) IsNotMountPoint(dir string) (bool, error)

func (*NsenterMounter) List added in v0.17.0 

func (*NsenterMounter) List() ([]MountPoint, error)

List returns a list of all mounted filesystems in the host's mount namespace.

func (*NsenterMounter) MakeRShared added in v1.8.0 

func (n *NsenterMounter) MakeRShared(path string) error

func (*NsenterMounter) Mount added in v0.17.0 

func (n *NsenterMounter) Mount(source string, target string, fstype string, options []string) error

Mount runs mount(8) in the host's root mount namespace. Aside from this aspect, Mount has the same semantics as the mounter returned by mount.New()

func (*NsenterMounter) PathIsDevice added in v1.4.0 

func (n *NsenterMounter) PathIsDevice(pathname string) (bool, error)

PathIsDevice uses FileInfo returned from os.Stat to check if path refers to a device.

func (*NsenterMounter) PrepareSafeSubpath added in v1.8.9 

func (mounter *NsenterMounter) PrepareSafeSubpath(subPath Subpath) (newHostPath string, cleanupAction func(), err error)

func (*NsenterMounter) SafeMakeDir added in v1.8.9 

func (mounter *NsenterMounter) SafeMakeDir(pathname string, base string, perm os.FileMode) error

func (*NsenterMounter) Unmount added in v0.17.0 

func (n *NsenterMounter) Unmount(target string) error

Unmount runs umount(8) in the host's mount namespace.

type SafeFormatAndMount added in v1.1.0 

type SafeFormatAndMount struct {
	Interface
	Exec
}

SafeFormatAndMount probes a device to see if it is formatted. Namely it checks to see if a file system is present. If so it mounts it otherwise the device is formatted first then mounted.

func (*SafeFormatAndMount) FormatAndMount added in v1.2.0 

func (mounter *SafeFormatAndMount) FormatAndMount(source string, target string, fstype string, options []string) error

FormatAndMount formats the given disk, if needed, and mounts it. That is if the disk is not formatted and it is not being mounted as read-only it will format it first then mount it. Otherwise, if the disk is already formatted or it is being mounted as read-only, it will be mounted without formatting.

type Subpath added in v1.8.9 

type Subpath struct {
	// index of the VolumeMount for this container
	VolumeMountIndex int
	// Full path to the subpath directory on the host
	Path string
	// name of the volume that is a valid directory name.
	VolumeName string
	// Full path to the volume path
	VolumePath string
	// Path to the pod's directory, including pod UID
	PodDir string
	// Name of the container
	ContainerName string
}

Source Files

View all Source files

Jump to

Keyboard shortcuts

? : This menu
/ : Search site
f or F : Jump to
y or Y : Canonical URL
go.dev uses cookies from Google to deliver and enhance the quality of its services and to analyze traffic. Learn more.