Windows: Experimental: Allow containerd for runtime
Signed-off-by: John Howard <jhoward@microsoft.com>
This is the first step in refactoring moby (dockerd) to use containerd on Windows.
Similar to the current model in Linux, this adds the option to enable it for runtime.
It does not switch the graphdriver to containerd snapshotters.
- Refactors libcontainerd to a series of subpackages so that either a
"local" containerd (1) or a "remote" (2) containerd can be loaded as opposed
to conditional compile as "local" for Windows and "remote" for Linux.
- Updates libcontainerd such that Windows has an option to allow the use of a
"remote" containerd. Here, it communicates over a named pipe using GRPC.
This is currently guarded behind the experimental flag, an environment variable,
and the providing of a pipename to connect to containerd.
- Infrastructure pieces such as under pkg/system to have helper functions for
determining whether containerd is being used.
(1) "local" containerd is what the daemon on Windows has used since inception.
It's not really containerd at all - it's simply local invocation of HCS APIs
directly in-process from the daemon through the Microsoft/hcsshim library.
(2) "remote" containerd is what docker on Linux uses for it's runtime. It means
that there is a separate containerd service running, and docker communicates over
GRPC to it.
To try this out, you will need to start with something like the following:
Window 1:
containerd --log-level debug
Window 2:
$env:DOCKER_WINDOWS_CONTAINERD=1
dockerd --experimental -D --containerd \\.\pipe\containerd-containerd
You will need the following binary from github.com/containerd/containerd in your path:
- containerd.exe
You will need the following binaries from github.com/Microsoft/hcsshim in your path:
- runhcs.exe
- containerd-shim-runhcs-v1.exe
For LCOW, it will require and initrd.img and kernel in `C:\Program Files\Linux Containers`.
This is no different to the current requirements. However, you may need updated binaries,
particularly initrd.img built from Microsoft/opengcs as (at the time of writing), Linuxkit
binaries are somewhat out of date.
Note that containerd and hcsshim for HCS v2 APIs do not yet support all the required
functionality needed for docker. This will come in time - this is a baby (although large)
step to migrating Docker on Windows to containerd.
Note that the HCS v2 APIs are only called on RS5+ builds. RS1..RS4 will still use
HCS v1 APIs as the v2 APIs were not fully developed enough on these builds to be usable.
This abstraction is done in HCSShim. (Referring specifically to runtime)
Note the LCOW graphdriver still uses HCS v1 APIs regardless.
Note also that this does not migrate docker to use containerd snapshotters
rather than graphdrivers. This needs to be done in conjunction with Linux also
doing the same switch.
2019-01-08 17:30:52 -05:00
|
|
|
package remote // import "github.com/docker/docker/libcontainerd/remote"
|
2017-09-22 09:52:41 -04:00
|
|
|
|
|
|
|
import (
|
|
|
|
"context"
|
|
|
|
"fmt"
|
|
|
|
"os"
|
|
|
|
"path/filepath"
|
|
|
|
"strings"
|
|
|
|
|
|
|
|
"github.com/containerd/containerd"
|
2017-11-29 19:15:20 -05:00
|
|
|
"github.com/containerd/containerd/cio"
|
2019-03-25 14:19:09 -04:00
|
|
|
"github.com/containerd/containerd/containers"
|
Windows: Experimental: Allow containerd for runtime
Signed-off-by: John Howard <jhoward@microsoft.com>
This is the first step in refactoring moby (dockerd) to use containerd on Windows.
Similar to the current model in Linux, this adds the option to enable it for runtime.
It does not switch the graphdriver to containerd snapshotters.
- Refactors libcontainerd to a series of subpackages so that either a
"local" containerd (1) or a "remote" (2) containerd can be loaded as opposed
to conditional compile as "local" for Windows and "remote" for Linux.
- Updates libcontainerd such that Windows has an option to allow the use of a
"remote" containerd. Here, it communicates over a named pipe using GRPC.
This is currently guarded behind the experimental flag, an environment variable,
and the providing of a pipename to connect to containerd.
- Infrastructure pieces such as under pkg/system to have helper functions for
determining whether containerd is being used.
(1) "local" containerd is what the daemon on Windows has used since inception.
It's not really containerd at all - it's simply local invocation of HCS APIs
directly in-process from the daemon through the Microsoft/hcsshim library.
(2) "remote" containerd is what docker on Linux uses for it's runtime. It means
that there is a separate containerd service running, and docker communicates over
GRPC to it.
To try this out, you will need to start with something like the following:
Window 1:
containerd --log-level debug
Window 2:
$env:DOCKER_WINDOWS_CONTAINERD=1
dockerd --experimental -D --containerd \\.\pipe\containerd-containerd
You will need the following binary from github.com/containerd/containerd in your path:
- containerd.exe
You will need the following binaries from github.com/Microsoft/hcsshim in your path:
- runhcs.exe
- containerd-shim-runhcs-v1.exe
For LCOW, it will require and initrd.img and kernel in `C:\Program Files\Linux Containers`.
This is no different to the current requirements. However, you may need updated binaries,
particularly initrd.img built from Microsoft/opengcs as (at the time of writing), Linuxkit
binaries are somewhat out of date.
Note that containerd and hcsshim for HCS v2 APIs do not yet support all the required
functionality needed for docker. This will come in time - this is a baby (although large)
step to migrating Docker on Windows to containerd.
Note that the HCS v2 APIs are only called on RS5+ builds. RS1..RS4 will still use
HCS v1 APIs as the v2 APIs were not fully developed enough on these builds to be usable.
This abstraction is done in HCSShim. (Referring specifically to runtime)
Note the LCOW graphdriver still uses HCS v1 APIs regardless.
Note also that this does not migrate docker to use containerd snapshotters
rather than graphdrivers. This needs to be done in conjunction with Linux also
doing the same switch.
2019-01-08 17:30:52 -05:00
|
|
|
libcontainerdtypes "github.com/docker/docker/libcontainerd/types"
|
2017-09-22 09:52:41 -04:00
|
|
|
"github.com/docker/docker/pkg/idtools"
|
2019-08-05 10:37:47 -04:00
|
|
|
specs "github.com/opencontainers/runtime-spec/specs-go"
|
2017-11-13 17:53:56 -05:00
|
|
|
"github.com/sirupsen/logrus"
|
2017-09-22 09:52:41 -04:00
|
|
|
)
|
|
|
|
|
Windows: Experimental: Allow containerd for runtime
Signed-off-by: John Howard <jhoward@microsoft.com>
This is the first step in refactoring moby (dockerd) to use containerd on Windows.
Similar to the current model in Linux, this adds the option to enable it for runtime.
It does not switch the graphdriver to containerd snapshotters.
- Refactors libcontainerd to a series of subpackages so that either a
"local" containerd (1) or a "remote" (2) containerd can be loaded as opposed
to conditional compile as "local" for Windows and "remote" for Linux.
- Updates libcontainerd such that Windows has an option to allow the use of a
"remote" containerd. Here, it communicates over a named pipe using GRPC.
This is currently guarded behind the experimental flag, an environment variable,
and the providing of a pipename to connect to containerd.
- Infrastructure pieces such as under pkg/system to have helper functions for
determining whether containerd is being used.
(1) "local" containerd is what the daemon on Windows has used since inception.
It's not really containerd at all - it's simply local invocation of HCS APIs
directly in-process from the daemon through the Microsoft/hcsshim library.
(2) "remote" containerd is what docker on Linux uses for it's runtime. It means
that there is a separate containerd service running, and docker communicates over
GRPC to it.
To try this out, you will need to start with something like the following:
Window 1:
containerd --log-level debug
Window 2:
$env:DOCKER_WINDOWS_CONTAINERD=1
dockerd --experimental -D --containerd \\.\pipe\containerd-containerd
You will need the following binary from github.com/containerd/containerd in your path:
- containerd.exe
You will need the following binaries from github.com/Microsoft/hcsshim in your path:
- runhcs.exe
- containerd-shim-runhcs-v1.exe
For LCOW, it will require and initrd.img and kernel in `C:\Program Files\Linux Containers`.
This is no different to the current requirements. However, you may need updated binaries,
particularly initrd.img built from Microsoft/opengcs as (at the time of writing), Linuxkit
binaries are somewhat out of date.
Note that containerd and hcsshim for HCS v2 APIs do not yet support all the required
functionality needed for docker. This will come in time - this is a baby (although large)
step to migrating Docker on Windows to containerd.
Note that the HCS v2 APIs are only called on RS5+ builds. RS1..RS4 will still use
HCS v1 APIs as the v2 APIs were not fully developed enough on these builds to be usable.
This abstraction is done in HCSShim. (Referring specifically to runtime)
Note the LCOW graphdriver still uses HCS v1 APIs regardless.
Note also that this does not migrate docker to use containerd snapshotters
rather than graphdrivers. This needs to be done in conjunction with Linux also
doing the same switch.
2019-01-08 17:30:52 -05:00
|
|
|
const runtimeName = "io.containerd.runtime.v1.linux"
|
|
|
|
|
|
|
|
func summaryFromInterface(i interface{}) (*libcontainerdtypes.Summary, error) {
|
|
|
|
return &libcontainerdtypes.Summary{}, nil
|
2017-09-22 09:52:41 -04:00
|
|
|
}
|
|
|
|
|
Windows: Experimental: Allow containerd for runtime
Signed-off-by: John Howard <jhoward@microsoft.com>
This is the first step in refactoring moby (dockerd) to use containerd on Windows.
Similar to the current model in Linux, this adds the option to enable it for runtime.
It does not switch the graphdriver to containerd snapshotters.
- Refactors libcontainerd to a series of subpackages so that either a
"local" containerd (1) or a "remote" (2) containerd can be loaded as opposed
to conditional compile as "local" for Windows and "remote" for Linux.
- Updates libcontainerd such that Windows has an option to allow the use of a
"remote" containerd. Here, it communicates over a named pipe using GRPC.
This is currently guarded behind the experimental flag, an environment variable,
and the providing of a pipename to connect to containerd.
- Infrastructure pieces such as under pkg/system to have helper functions for
determining whether containerd is being used.
(1) "local" containerd is what the daemon on Windows has used since inception.
It's not really containerd at all - it's simply local invocation of HCS APIs
directly in-process from the daemon through the Microsoft/hcsshim library.
(2) "remote" containerd is what docker on Linux uses for it's runtime. It means
that there is a separate containerd service running, and docker communicates over
GRPC to it.
To try this out, you will need to start with something like the following:
Window 1:
containerd --log-level debug
Window 2:
$env:DOCKER_WINDOWS_CONTAINERD=1
dockerd --experimental -D --containerd \\.\pipe\containerd-containerd
You will need the following binary from github.com/containerd/containerd in your path:
- containerd.exe
You will need the following binaries from github.com/Microsoft/hcsshim in your path:
- runhcs.exe
- containerd-shim-runhcs-v1.exe
For LCOW, it will require and initrd.img and kernel in `C:\Program Files\Linux Containers`.
This is no different to the current requirements. However, you may need updated binaries,
particularly initrd.img built from Microsoft/opengcs as (at the time of writing), Linuxkit
binaries are somewhat out of date.
Note that containerd and hcsshim for HCS v2 APIs do not yet support all the required
functionality needed for docker. This will come in time - this is a baby (although large)
step to migrating Docker on Windows to containerd.
Note that the HCS v2 APIs are only called on RS5+ builds. RS1..RS4 will still use
HCS v1 APIs as the v2 APIs were not fully developed enough on these builds to be usable.
This abstraction is done in HCSShim. (Referring specifically to runtime)
Note the LCOW graphdriver still uses HCS v1 APIs regardless.
Note also that this does not migrate docker to use containerd snapshotters
rather than graphdrivers. This needs to be done in conjunction with Linux also
doing the same switch.
2019-01-08 17:30:52 -05:00
|
|
|
func (c *client) UpdateResources(ctx context.Context, containerID string, resources *libcontainerdtypes.Resources) error {
|
2019-03-25 16:17:17 -04:00
|
|
|
p, err := c.getProcess(ctx, containerID, libcontainerdtypes.InitProcessName)
|
2017-09-22 09:52:41 -04:00
|
|
|
if err != nil {
|
|
|
|
return err
|
|
|
|
}
|
|
|
|
|
|
|
|
// go doesn't like the alias in 1.8, this means this need to be
|
|
|
|
// platform specific
|
|
|
|
return p.(containerd.Task).Update(ctx, containerd.WithResources((*specs.LinuxResources)(resources)))
|
|
|
|
}
|
|
|
|
|
|
|
|
func hostIDFromMap(id uint32, mp []specs.LinuxIDMapping) int {
|
|
|
|
for _, m := range mp {
|
|
|
|
if id >= m.ContainerID && id <= m.ContainerID+m.Size-1 {
|
|
|
|
return int(m.HostID + id - m.ContainerID)
|
|
|
|
}
|
|
|
|
}
|
|
|
|
return 0
|
|
|
|
}
|
|
|
|
|
|
|
|
func getSpecUser(ociSpec *specs.Spec) (int, int) {
|
|
|
|
var (
|
|
|
|
uid int
|
|
|
|
gid int
|
|
|
|
)
|
|
|
|
|
|
|
|
for _, ns := range ociSpec.Linux.Namespaces {
|
|
|
|
if ns.Type == specs.UserNamespace {
|
|
|
|
uid = hostIDFromMap(0, ociSpec.Linux.UIDMappings)
|
|
|
|
gid = hostIDFromMap(0, ociSpec.Linux.GIDMappings)
|
|
|
|
break
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
return uid, gid
|
|
|
|
}
|
|
|
|
|
2019-03-25 14:19:09 -04:00
|
|
|
// WithBundle creates the bundle for the container
|
|
|
|
func WithBundle(bundleDir string, ociSpec *specs.Spec) containerd.NewContainerOpts {
|
|
|
|
return func(ctx context.Context, client *containerd.Client, c *containers.Container) error {
|
2019-03-25 16:17:17 -04:00
|
|
|
if c.Labels == nil {
|
|
|
|
c.Labels = make(map[string]string)
|
|
|
|
}
|
2019-03-25 14:19:09 -04:00
|
|
|
uid, gid := getSpecUser(ociSpec)
|
|
|
|
if uid == 0 && gid == 0 {
|
2019-03-25 16:17:17 -04:00
|
|
|
c.Labels[DockerContainerBundlePath] = bundleDir
|
2019-03-25 14:19:09 -04:00
|
|
|
return idtools.MkdirAllAndChownNew(bundleDir, 0755, idtools.Identity{UID: 0, GID: 0})
|
2017-09-22 09:52:41 -04:00
|
|
|
}
|
2019-03-25 14:19:09 -04:00
|
|
|
|
|
|
|
p := string(filepath.Separator)
|
|
|
|
components := strings.Split(bundleDir, string(filepath.Separator))
|
|
|
|
for _, d := range components[1:] {
|
|
|
|
p = filepath.Join(p, d)
|
|
|
|
fi, err := os.Stat(p)
|
|
|
|
if err != nil && !os.IsNotExist(err) {
|
|
|
|
return err
|
|
|
|
}
|
|
|
|
if os.IsNotExist(err) || fi.Mode()&1 == 0 {
|
|
|
|
p = fmt.Sprintf("%s.%d.%d", p, uid, gid)
|
|
|
|
if err := idtools.MkdirAndChown(p, 0700, idtools.Identity{UID: uid, GID: gid}); err != nil && !os.IsExist(err) {
|
|
|
|
return err
|
|
|
|
}
|
2017-09-22 09:52:41 -04:00
|
|
|
}
|
|
|
|
}
|
2019-03-25 16:17:17 -04:00
|
|
|
if c.Labels == nil {
|
|
|
|
c.Labels = make(map[string]string)
|
|
|
|
}
|
|
|
|
c.Labels[DockerContainerBundlePath] = p
|
2019-03-25 14:19:09 -04:00
|
|
|
return nil
|
2017-09-22 09:52:41 -04:00
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2017-12-07 14:26:27 -05:00
|
|
|
func newFIFOSet(bundleDir, processID string, withStdin, withTerminal bool) *cio.FIFOSet {
|
2017-12-07 15:52:14 -05:00
|
|
|
config := cio.Config{
|
|
|
|
Terminal: withTerminal,
|
|
|
|
Stdout: filepath.Join(bundleDir, processID+"-stdout"),
|
2017-09-22 09:52:41 -04:00
|
|
|
}
|
2017-12-07 15:52:14 -05:00
|
|
|
paths := []string{config.Stdout}
|
2017-09-22 09:52:41 -04:00
|
|
|
|
|
|
|
if withStdin {
|
2017-12-07 15:52:14 -05:00
|
|
|
config.Stdin = filepath.Join(bundleDir, processID+"-stdin")
|
|
|
|
paths = append(paths, config.Stdin)
|
2017-09-22 09:52:41 -04:00
|
|
|
}
|
2017-12-07 15:52:14 -05:00
|
|
|
if !withTerminal {
|
|
|
|
config.Stderr = filepath.Join(bundleDir, processID+"-stderr")
|
|
|
|
paths = append(paths, config.Stderr)
|
2017-09-22 09:52:41 -04:00
|
|
|
}
|
2017-12-07 15:52:14 -05:00
|
|
|
closer := func() error {
|
|
|
|
for _, path := range paths {
|
|
|
|
if err := os.RemoveAll(path); err != nil {
|
|
|
|
logrus.Warnf("libcontainerd: failed to remove fifo %v: %v", path, err)
|
2017-11-13 17:53:56 -05:00
|
|
|
}
|
|
|
|
}
|
2017-12-07 15:52:14 -05:00
|
|
|
return nil
|
2017-11-13 17:53:56 -05:00
|
|
|
}
|
2017-12-07 15:52:14 -05:00
|
|
|
|
|
|
|
return cio.NewFIFOSet(config, closer)
|
2017-11-13 17:53:56 -05:00
|
|
|
}
|
Windows: Experimental: Allow containerd for runtime
Signed-off-by: John Howard <jhoward@microsoft.com>
This is the first step in refactoring moby (dockerd) to use containerd on Windows.
Similar to the current model in Linux, this adds the option to enable it for runtime.
It does not switch the graphdriver to containerd snapshotters.
- Refactors libcontainerd to a series of subpackages so that either a
"local" containerd (1) or a "remote" (2) containerd can be loaded as opposed
to conditional compile as "local" for Windows and "remote" for Linux.
- Updates libcontainerd such that Windows has an option to allow the use of a
"remote" containerd. Here, it communicates over a named pipe using GRPC.
This is currently guarded behind the experimental flag, an environment variable,
and the providing of a pipename to connect to containerd.
- Infrastructure pieces such as under pkg/system to have helper functions for
determining whether containerd is being used.
(1) "local" containerd is what the daemon on Windows has used since inception.
It's not really containerd at all - it's simply local invocation of HCS APIs
directly in-process from the daemon through the Microsoft/hcsshim library.
(2) "remote" containerd is what docker on Linux uses for it's runtime. It means
that there is a separate containerd service running, and docker communicates over
GRPC to it.
To try this out, you will need to start with something like the following:
Window 1:
containerd --log-level debug
Window 2:
$env:DOCKER_WINDOWS_CONTAINERD=1
dockerd --experimental -D --containerd \\.\pipe\containerd-containerd
You will need the following binary from github.com/containerd/containerd in your path:
- containerd.exe
You will need the following binaries from github.com/Microsoft/hcsshim in your path:
- runhcs.exe
- containerd-shim-runhcs-v1.exe
For LCOW, it will require and initrd.img and kernel in `C:\Program Files\Linux Containers`.
This is no different to the current requirements. However, you may need updated binaries,
particularly initrd.img built from Microsoft/opengcs as (at the time of writing), Linuxkit
binaries are somewhat out of date.
Note that containerd and hcsshim for HCS v2 APIs do not yet support all the required
functionality needed for docker. This will come in time - this is a baby (although large)
step to migrating Docker on Windows to containerd.
Note that the HCS v2 APIs are only called on RS5+ builds. RS1..RS4 will still use
HCS v1 APIs as the v2 APIs were not fully developed enough on these builds to be usable.
This abstraction is done in HCSShim. (Referring specifically to runtime)
Note the LCOW graphdriver still uses HCS v1 APIs regardless.
Note also that this does not migrate docker to use containerd snapshotters
rather than graphdrivers. This needs to be done in conjunction with Linux also
doing the same switch.
2019-01-08 17:30:52 -05:00
|
|
|
|
|
|
|
func (c *client) newDirectIO(ctx context.Context, fifos *cio.FIFOSet) (*cio.DirectIO, error) {
|
|
|
|
return cio.NewDirectIO(ctx, fifos)
|
|
|
|
}
|