2017-11-01 19:37:53 -04:00
|
|
|
// +build !windows
|
2015-05-15 19:34:26 -04:00
|
|
|
|
|
|
|
package main
|
|
|
|
|
|
|
|
import (
|
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binaries are somewhat out of date.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
|
|
|
"context"
|
2015-06-15 09:36:19 -04:00
|
|
|
"fmt"
|
2016-04-09 02:49:33 -04:00
|
|
|
"net"
|
2015-05-15 19:34:26 -04:00
|
|
|
"os"
|
2015-12-10 18:35:10 -05:00
|
|
|
"os/signal"
|
2016-03-24 14:42:03 -04:00
|
|
|
"path/filepath"
|
2016-04-09 02:49:33 -04:00
|
|
|
"strconv"
|
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
|
|
|
"time"
|
2015-05-15 19:34:26 -04:00
|
|
|
|
2018-08-17 16:08:22 -04:00
|
|
|
"github.com/containerd/containerd/runtime/v1/linux"
|
2016-04-12 11:21:20 -04:00
|
|
|
"github.com/docker/docker/cmd/dockerd/hack"
|
2015-05-15 19:34:26 -04:00
|
|
|
"github.com/docker/docker/daemon"
|
2018-09-07 16:43:21 -04:00
|
|
|
"github.com/docker/docker/daemon/config"
|
2018-05-23 15:15:21 -04:00
|
|
|
"github.com/docker/docker/libcontainerd/supervisor"
|
2018-10-15 03:52:53 -04:00
|
|
|
"github.com/docker/docker/pkg/homedir"
|
2016-04-09 02:49:33 -04:00
|
|
|
"github.com/docker/libnetwork/portallocator"
|
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
|
|
|
"github.com/pkg/errors"
|
2017-05-23 10:22:32 -04:00
|
|
|
"golang.org/x/sys/unix"
|
2015-05-15 19:34:26 -04:00
|
|
|
)
|
|
|
|
|
2018-10-15 03:52:53 -04:00
|
|
|
func getDefaultDaemonConfigDir() (string, error) {
|
2019-04-19 03:53:58 -04:00
|
|
|
if !honorXDG {
|
2018-10-15 03:52:53 -04:00
|
|
|
return "/etc/docker", nil
|
|
|
|
}
|
|
|
|
// NOTE: CLI uses ~/.docker while the daemon uses ~/.config/docker, because
|
|
|
|
// ~/.docker was not designed to store daemon configurations.
|
|
|
|
// In future, the daemon directory may be renamed to ~/.config/moby-engine (?).
|
|
|
|
configHome, err := homedir.GetConfigHome()
|
|
|
|
if err != nil {
|
|
|
|
return "", nil
|
|
|
|
}
|
|
|
|
return filepath.Join(configHome, "docker"), nil
|
|
|
|
}
|
|
|
|
|
|
|
|
func getDefaultDaemonConfigFile() (string, error) {
|
|
|
|
dir, err := getDefaultDaemonConfigDir()
|
|
|
|
if err != nil {
|
|
|
|
return "", err
|
|
|
|
}
|
|
|
|
return filepath.Join(dir, "daemon.json"), nil
|
|
|
|
}
|
2015-12-10 18:35:10 -05:00
|
|
|
|
2015-06-15 09:36:19 -04:00
|
|
|
// setDefaultUmask sets the umask to 0022 to avoid problems
|
|
|
|
// caused by custom umask
|
|
|
|
func setDefaultUmask() error {
|
|
|
|
desiredUmask := 0022
|
2017-05-23 10:22:32 -04:00
|
|
|
unix.Umask(desiredUmask)
|
|
|
|
if umask := unix.Umask(desiredUmask); umask != desiredUmask {
|
2015-06-15 09:36:19 -04:00
|
|
|
return fmt.Errorf("failed to set umask: expected %#o, got %#o", desiredUmask, umask)
|
|
|
|
}
|
|
|
|
|
|
|
|
return nil
|
|
|
|
}
|
2015-07-30 15:39:19 -04:00
|
|
|
|
2018-05-23 15:15:21 -04:00
|
|
|
func (cli *DaemonCli) getPlatformContainerdDaemonOpts() ([]supervisor.DaemonOpt, error) {
|
|
|
|
opts := []supervisor.DaemonOpt{
|
|
|
|
supervisor.WithOOMScore(cli.Config.OOMScoreAdjust),
|
|
|
|
supervisor.WithPlugin("linux", &linux.Config{
|
2017-12-04 14:17:30 -05:00
|
|
|
Shim: daemon.DefaultShimBinary,
|
|
|
|
Runtime: daemon.DefaultRuntimeBinary,
|
|
|
|
RuntimeRoot: filepath.Join(cli.Config.Root, "runc"),
|
|
|
|
ShimDebug: cli.Config.Debug,
|
2017-09-22 09:52:41 -04:00
|
|
|
}),
|
|
|
|
}
|
|
|
|
|
|
|
|
return opts, nil
|
2016-03-18 14:50:19 -04:00
|
|
|
}
|
2016-03-24 14:42:03 -04:00
|
|
|
|
2017-09-22 09:52:41 -04:00
|
|
|
// setupConfigReloadTrap configures the USR2 signal to reload the configuration.
|
|
|
|
func (cli *DaemonCli) setupConfigReloadTrap() {
|
|
|
|
c := make(chan os.Signal, 1)
|
|
|
|
signal.Notify(c, unix.SIGHUP)
|
|
|
|
go func() {
|
|
|
|
for range c {
|
|
|
|
cli.reloadConfig()
|
|
|
|
}
|
|
|
|
}()
|
2016-03-24 14:42:03 -04:00
|
|
|
}
|
2016-04-09 02:49:33 -04:00
|
|
|
|
2016-08-19 16:06:28 -04:00
|
|
|
// getSwarmRunRoot gets the root directory for swarm to store runtime state
|
|
|
|
// For example, the control socket
|
|
|
|
func (cli *DaemonCli) getSwarmRunRoot() string {
|
|
|
|
return filepath.Join(cli.Config.ExecRoot, "swarm")
|
|
|
|
}
|
|
|
|
|
2016-04-09 02:49:33 -04:00
|
|
|
// allocateDaemonPort ensures that there are no containers
|
|
|
|
// that try to use any port allocated for the docker server.
|
|
|
|
func allocateDaemonPort(addr string) error {
|
|
|
|
host, port, err := net.SplitHostPort(addr)
|
|
|
|
if err != nil {
|
|
|
|
return err
|
|
|
|
}
|
|
|
|
|
|
|
|
intPort, err := strconv.Atoi(port)
|
|
|
|
if err != nil {
|
|
|
|
return err
|
|
|
|
}
|
|
|
|
|
|
|
|
var hostIPs []net.IP
|
|
|
|
if parsedIP := net.ParseIP(host); parsedIP != nil {
|
|
|
|
hostIPs = append(hostIPs, parsedIP)
|
|
|
|
} else if hostIPs, err = net.LookupIP(host); err != nil {
|
|
|
|
return fmt.Errorf("failed to lookup %s address in host specification", host)
|
|
|
|
}
|
|
|
|
|
|
|
|
pa := portallocator.Get()
|
|
|
|
for _, hostIP := range hostIPs {
|
|
|
|
if _, err := pa.RequestPort(hostIP, "tcp", intPort); err != nil {
|
|
|
|
return fmt.Errorf("failed to allocate daemon listening port %d (err: %v)", intPort, err)
|
|
|
|
}
|
|
|
|
}
|
|
|
|
return nil
|
|
|
|
}
|
2016-04-22 20:16:14 -04:00
|
|
|
|
2016-04-12 11:21:20 -04:00
|
|
|
func wrapListeners(proto string, ls []net.Listener) []net.Listener {
|
2016-05-21 23:40:15 -04:00
|
|
|
switch proto {
|
|
|
|
case "unix":
|
2018-05-17 10:29:52 -04:00
|
|
|
ls[0] = &hack.MalformedHostHeaderOverride{Listener: ls[0]}
|
2016-05-21 23:40:15 -04:00
|
|
|
case "fd":
|
|
|
|
for i := range ls {
|
2018-05-17 10:29:52 -04:00
|
|
|
ls[i] = &hack.MalformedHostHeaderOverride{Listener: ls[i]}
|
2016-04-12 11:21:20 -04:00
|
|
|
}
|
|
|
|
}
|
|
|
|
return ls
|
|
|
|
}
|
2018-09-07 16:43:21 -04:00
|
|
|
|
|
|
|
func newCgroupParent(config *config.Config) string {
|
|
|
|
cgroupParent := "docker"
|
|
|
|
useSystemd := daemon.UsingSystemd(config)
|
|
|
|
if useSystemd {
|
|
|
|
cgroupParent = "system.slice"
|
|
|
|
}
|
|
|
|
if config.CgroupParent != "" {
|
|
|
|
cgroupParent = config.CgroupParent
|
|
|
|
}
|
|
|
|
if useSystemd {
|
|
|
|
cgroupParent = cgroupParent + ":" + "docker" + ":"
|
|
|
|
}
|
|
|
|
return cgroupParent
|
|
|
|
}
|
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
|
|
|
|
2019-04-08 05:44:14 -04:00
|
|
|
func (cli *DaemonCli) initContainerD(ctx context.Context) (func(time.Duration) error, error) {
|
|
|
|
var waitForShutdown func(time.Duration) error
|
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
|
|
|
if cli.Config.ContainerdAddr == "" {
|
2019-04-19 03:53:58 -04:00
|
|
|
systemContainerdAddr, ok, err := systemContainerdRunning(honorXDG)
|
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
|
|
|
if err != nil {
|
2019-04-08 05:44:14 -04:00
|
|
|
return nil, errors.Wrap(err, "could not determine whether the system containerd is running")
|
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
|
|
|
}
|
|
|
|
if !ok {
|
|
|
|
opts, err := cli.getContainerdDaemonOpts()
|
|
|
|
if err != nil {
|
2019-04-08 05:44:14 -04:00
|
|
|
return nil, errors.Wrap(err, "failed to generate containerd options")
|
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
|
|
|
}
|
|
|
|
|
|
|
|
r, err := supervisor.Start(ctx, filepath.Join(cli.Config.Root, "containerd"), filepath.Join(cli.Config.ExecRoot, "containerd"), opts...)
|
|
|
|
if err != nil {
|
2019-04-08 05:44:14 -04:00
|
|
|
return nil, errors.Wrap(err, "failed to start containerd")
|
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
|
|
|
}
|
|
|
|
cli.Config.ContainerdAddr = r.Address()
|
|
|
|
|
|
|
|
// Try to wait for containerd to shutdown
|
2019-04-08 05:44:14 -04:00
|
|
|
waitForShutdown = r.WaitTimeout
|
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
|
|
|
} else {
|
|
|
|
cli.Config.ContainerdAddr = systemContainerdAddr
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2019-04-08 05:44:14 -04:00
|
|
|
return waitForShutdown, nil
|
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
|
|
|
}
|