We need to isolate the images that we are remapping to a userns, we
can't mix them with "normal" images. In the graph driver case this means
we create a new root directory where we store the images and everything
else, in the containerd case we can use a new namespace.
Signed-off-by: Djordje Lukic <djordje.lukic@docker.com>
- pass the cluster as an argument instead of manually setting it after
creating the router-options
- remove the "opts" variable, to prevent it accidentally being used (with
the assumption that's the value returned)
- use a struct-literal for the returned options.
Signed-off-by: Sebastiaan van Stijn <github@gone.nl>
Commit 21e50b89c9 added a label on the buildkit
worker to advertise the host-gateway-ip. This option can be either set by the
user in the daemon config, or otherwise defaults to the gateway-ip.
If no value is set by the user, discovery of the gateway-ip happens when
initializing the network-controller (`NewDaemon`, `daemon.restore()`).
However d222bf097c changed how we handle the
daemon config. As a result, the `cli.Config` used when initializing the
builder only holds configuration information form the daemon config
(user-specified or defaults), but is not updated with information set
by `NewDaemon`.
This patch adds an accessor on the daemon to get the current daemon config.
An alternative could be to return the config by `NewDaemon` (which should
likely be a _copy_ of the config).
Before this patch:
docker buildx inspect default
Name: default
Driver: docker
Nodes:
Name: default
Endpoint: default
Status: running
Buildkit: v0.12.4+3b6880d2a00f
Platforms: linux/arm64, linux/amd64, linux/amd64/v2, linux/riscv64, linux/ppc64le, linux/s390x, linux/386, linux/mips64le, linux/mips64, linux/arm/v7, linux/arm/v6
Labels:
org.mobyproject.buildkit.worker.moby.host-gateway-ip: <nil>
After this patch:
docker buildx inspect default
Name: default
Driver: docker
Nodes:
Name: default
Endpoint: default
Status: running
Buildkit: v0.12.4+3b6880d2a00f
Platforms: linux/arm64, linux/amd64, linux/amd64/v2, linux/riscv64, linux/ppc64le, linux/s390x, linux/386, linux/mips64le, linux/mips64, linux/arm/v7, linux/arm/v6
Labels:
org.mobyproject.buildkit.worker.moby.host-gateway-ip: 172.18.0.1
Signed-off-by: Sebastiaan van Stijn <github@gone.nl>
This flag was marked deprecated in commit 5a922dc16 (released in v24.0)
and to be removed in the next release.
Signed-off-by: Albin Kerouanton <albinker@gmail.com>
commit 4f9db655ed moved looking up the
userland-proxy binary to early in the startup process, and introduced
a log-message if the binary was missing.
However, a side-effect of this was this message would also be printed
when running "--version";
dockerd --version
time="2024-01-09T09:18:53.705271292Z" level=warning msg="failed to lookup default userland-proxy binary" error="exec: \"docker-proxy\": executable file not found in $PATH"
Docker version v25.0.0-rc.1, build 9cebefa717
We should look if we can avoid this, but let's change the message to be
a debug message as a short-term workaround.
Signed-off-by: Sebastiaan van Stijn <github@gone.nl>
This commit switches our code to use semconv 1.21, which is the version matching
the OTEL modules, as well as the containerd code.
The BuildKit 0.12.x module currently uses an older version of the OTEL modules,
and uses the semconv 0.17 schema. Mixing schema-versions is problematic, but
we still want to consume BuildKit's "detect" package to wire-up other parts
of OTEL.
To align the versions in our code, this patch sets the BuildKit detect.Resource
with the correct semconv version.
It's worth noting that the BuildKit package has a custom "serviceNameDetector";
https://github.com/moby/buildkit/blob/v0.12.4/util/tracing/detect/detect.go#L153-L169
Whith is merged with OTEL's default resource:
https://github.com/moby/buildkit/blob/v0.12.4/util/tracing/detect/detect.go#L100-L107
There's no need to duplicate that code, as OTEL's `resource.Default()` already
provides this functionality:
- It uses fromEnv{} detector internally: https://github.com/open-telemetry/opentelemetry-go/blob/v1.19.0/sdk/resource/resource.go#L208
- fromEnv{} detector reads OTEL_SERVICE_NAME: https://github.com/open-telemetry/opentelemetry-go/blob/v1.19.0/sdk/resource/env.go#L53
This patch also removes uses of the httpconv package, which is no longer included
in semconv 1.21 and now an internal package. Removing the use of this package
means that hijacked connections will not have the HTTP attributes on the Moby
client span, which isn't ideal, but a limited loss that'd impact exec/attach.
The span itself will still exist, it just won't the additional attributes that
are added by that package.
Alternatively, the httpconv call COULD remain - it will not error and will send
syntactically valid spans but we would be mixing & matching semconv versions,
so won't be compliant.
Some parts of the httpconv package were preserved through a very minimal local
implementation; a variant of `httpconv.ClientStatus(resp.StatusCode))` is added
to set the span status (`span.SetStatus()`). The `httpconv` package has complex
logic for this, but mostly drills down to HTTP status range (1xx/2xx/3xx/4xx/5xx)
to determine if the status was successfull or non-successful (4xx/5xx).
The additional logic it provided was to validate actual status-codes, and to
convert "bogus" status codes in "success" ranges (1xx, 2xx) into an error. That
code seemed over-reaching (and not accounting for potential future _valid_
status codes). Let's assume we only get valid status codes.
- https://github.com/open-telemetry/opentelemetry-go/blob/v1.21.0/semconv/v1.17.0/httpconv/http.go#L85-L89
- https://github.com/open-telemetry/opentelemetry-go/blob/v1.21.0/semconv/internal/v2/http.go#L322-L330
- https://github.com/open-telemetry/opentelemetry-go/blob/v1.21.0/semconv/internal/v2/http.go#L356-L404
Signed-off-by: Sebastiaan van Stijn <github@gone.nl>
The daemon currently provides support for API versions all the way back
to v1.12, which is the version of the API that shipped with docker 1.0. On
Windows, the minimum supported version is v1.24.
Such old versions of the client are rare, and supporting older API versions
has accumulated significant amounts of code to remain backward-compatible
(which is largely untested, and a "best-effort" at most).
This patch updates the minimum API version to v1.24, which is the fallback
API version used when API-version negotiation fails. The intent is to start
deprecating older API versions, but no code is removed yet as part of this
patch, and a DOCKER_MIN_API_VERSION environment variable is added, which
allows overriding the minimum version (to allow restoring the behavior from
before this patch).
With this patch the daemon defaults to API v1.24 as minimum:
docker version
Client:
Version: 24.0.2
API version: 1.43
Go version: go1.20.4
Git commit: cb74dfc
Built: Thu May 25 21:50:49 2023
OS/Arch: linux/arm64
Context: default
Server:
Engine:
Version: dev
API version: 1.44 (minimum version 1.24)
Go version: go1.21.3
Git commit: 0322a29b9ef8806aaa4b45dc9d9a2ebcf0244bf4
Built: Mon Dec 4 15:22:17 2023
OS/Arch: linux/arm64
Experimental: false
containerd:
Version: v1.7.9
GitCommit: 4f03e100cb967922bec7459a78d16ccbac9bb81d
runc:
Version: 1.1.10
GitCommit: v1.1.10-0-g18a0cb0
docker-init:
Version: 0.19.0
GitCommit: de40ad0
Trying to use an older version of the API produces an error:
DOCKER_API_VERSION=1.23 docker version
Client:
Version: 24.0.2
API version: 1.23 (downgraded from 1.43)
Go version: go1.20.4
Git commit: cb74dfc
Built: Thu May 25 21:50:49 2023
OS/Arch: linux/arm64
Context: default
Error response from daemon: client version 1.23 is too old. Minimum supported API version is 1.24, please upgrade your client to a newer version
To restore the previous minimum, users can start the daemon with the
DOCKER_MIN_API_VERSION environment variable set:
DOCKER_MIN_API_VERSION=1.12 dockerd
API 1.12 is the oldest supported API version on Linux;
docker version
Client:
Version: 24.0.2
API version: 1.43
Go version: go1.20.4
Git commit: cb74dfc
Built: Thu May 25 21:50:49 2023
OS/Arch: linux/arm64
Context: default
Server:
Engine:
Version: dev
API version: 1.44 (minimum version 1.12)
Go version: go1.21.3
Git commit: 0322a29b9ef8806aaa4b45dc9d9a2ebcf0244bf4
Built: Mon Dec 4 15:22:17 2023
OS/Arch: linux/arm64
Experimental: false
containerd:
Version: v1.7.9
GitCommit: 4f03e100cb967922bec7459a78d16ccbac9bb81d
runc:
Version: 1.1.10
GitCommit: v1.1.10-0-g18a0cb0
docker-init:
Version: 0.19.0
GitCommit: de40ad0
When using the `DOCKER_MIN_API_VERSION` with a version of the API that
is not supported, an error is produced when starting the daemon;
DOCKER_MIN_API_VERSION=1.11 dockerd --validate
invalid DOCKER_MIN_API_VERSION: minimum supported API version is 1.12: 1.11
DOCKER_MIN_API_VERSION=1.45 dockerd --validate
invalid DOCKER_MIN_API_VERSION: maximum supported API version is 1.44: 1.45
Specifying a malformed API version also produces the same error;
DOCKER_MIN_API_VERSION=hello dockerd --validate
invalid DOCKER_MIN_API_VERSION: minimum supported API version is 1.12: hello
Signed-off-by: Sebastiaan van Stijn <github@gone.nl>
This removes various templating functions that were added for the
docker CLI. These are not needed for the dockerd binary, which does
not have subcommands or management commands.
Revert "Only hide commands if the env variable is set."
This reverts commit a7c8bcac2b.
Signed-off-by: Sebastiaan van Stijn <github@gone.nl>
Use a strong type for the DNS IP-addresses so that we can use flags.IPSliceVar,
instead of implementing our own option-type and validation.
Behavior should be the same, although error-messages have slightly changed:
Before this patch:
dockerd --dns 1.1.1.1oooo --validate
Status: invalid argument "1.1.1.1oooo" for "--dns" flag: 1.1.1.1oooo is not an ip address
See 'dockerd --help'., Code: 125
cat /etc/docker/daemon.json
{"dns": ["1.1.1.1"]}
dockerd --dns 2.2.2.2 --validate
unable to configure the Docker daemon with file /etc/docker/daemon.json: the following directives are specified both as a flag and in the configuration file: dns: (from flag: [2.2.2.2], from file: [1.1.1.1])
cat /etc/docker/daemon.json
{"dns": ["1.1.1.1oooo"]}
dockerd --validate
unable to configure the Docker daemon with file /etc/docker/daemon.json: merged configuration validation from file and command line flags failed: 1.1.1.1ooooo is not an ip address
With this patch:
dockerd --dns 1.1.1.1oooo --validate
Status: invalid argument "1.1.1.1oooo" for "--dns" flag: invalid string being converted to IP address: 1.1.1.1oooo
See 'dockerd --help'., Code: 125
cat /etc/docker/daemon.json
{"dns": ["1.1.1.1"]}
dockerd --dns 2.2.2.2 --validate
unable to configure the Docker daemon with file /etc/docker/daemon.json: the following directives are specified both as a flag and in the configuration file: dns: (from flag: [2.2.2.2], from file: [1.1.1.1])
cat /etc/docker/daemon.json
{"dns": ["1.1.1.1oooo"]}
dockerd --validate
unable to configure the Docker daemon with file /etc/docker/daemon.json: invalid IP address: 1.1.1.1oooo
Signed-off-by: Sebastiaan van Stijn <github@gone.nl>
update buildkit to the latest code in the v0.12 branch:
full diff: f94ed7cec3...6560bb937e
Signed-off-by: Sebastiaan van Stijn <github@gone.nl>
Signed-off-by: Tonis Tiigi <tonistiigi@gmail.com>
The github.com/containerd/containerd/log package was moved to a separate
module, which will also be used by upcoming (patch) releases of containerd.
This patch moves our own uses of the package to use the new module.
Signed-off-by: Sebastiaan van Stijn <github@gone.nl>
This was mistakenly added to bklog.
Since this is getting attached to the standard logger, and bklog is
using the standard logger, we only need this added once.
Signed-off-by: Brian Goff <cpuguy83@gmail.com>
This type was introduced in
0a79e67e4f
Make use of it throughout our log-format handling code, and convert back
to a string before we pass it to the containerd client.
Signed-off-by: Bjorn Neergaard <bjorn.neergaard@docker.com>
full diff: https://github.com/containerd/containerd/compare/v1.6.22...v1.6.24
v1.6.24 release notes:
full diff: https://github.com/containerd/containerd/compare/v1.6.23...v1.6.24
The twenty-fourth patch release for containerd 1.6 contains various fixes
and updates.
Notable Updates
- CRI: fix leaked shim caused by high IO pressure
- Update to go1.20.8
- Update runc to v1.1.9
- Backport: add configurable mount options to overlay snapshotter
- log: cleanups and improvements to decouple more from logrus
v1.6.23 release notes:
full diff: https://github.com/containerd/containerd/compare/v1.6.22...v1.6.23
The twenty-third patch release for containerd 1.6 contains various fixes
and updates.
Notable Updates
- Add stable ABI support in windows platform matcher + update hcsshim tag
- cri: Don't use rel path for image volumes
- Upgrade GitHub actions packages in release workflow
- update to go1.19.12
- backport: ro option for userxattr mount check + cherry-pick: Fix ro mount option being passed
Signed-off-by: Sebastiaan van Stijn <github@gone.nl>
This uses otel standard environment variables to configure tracing in
the daemon.
It also adds support for propagating trace contexts in the client and
reading those from the API server.
See
https://opentelemetry.io/docs/specs/otel/configuration/sdk-environment-variables/
for details on otel environment variables.
Signed-off-by: Brian Goff <cpuguy83@gmail.com>
Go 1.15.7 contained a security fix for CVE-2021-3115, which allowed arbitrary
code to be executed at build time when using cgo on Windows.
This issue was not limited to the go command itself, and could also affect binaries
that use `os.Command`, `os.LookPath`, etc.
From the related blogpost (https://blog.golang.org/path-security):
> Are your own programs affected?
>
> If you use exec.LookPath or exec.Command in your own programs, you only need to
> be concerned if you (or your users) run your program in a directory with untrusted
> contents. If so, then a subprocess could be started using an executable from dot
> instead of from a system directory. (Again, using an executable from dot happens
> always on Windows and only with uncommon PATH settings on Unix.)
>
> If you are concerned, then we’ve published the more restricted variant of os/exec
> as golang.org/x/sys/execabs. You can use it in your program by simply replacing
At time of the go1.15 release, the Go team considered changing the behavior of
`os.LookPath()` and `exec.LookPath()` to be a breaking change, and made the
behavior "opt-in" by providing the `golang.org/x/sys/execabs` package as a
replacement.
However, for the go1.19 release, this changed, and the default behavior of
`os.LookPath()` and `exec.LookPath()` was changed. From the release notes:
https://go.dev/doc/go1.19#os-exec-path
> Command and LookPath no longer allow results from a PATH search to be found
> relative to the current directory. This removes a common source of security
> problems but may also break existing programs that depend on using, say,
> exec.Command("prog") to run a binary named prog (or, on Windows, prog.exe)
> in the current directory. See the os/exec package documentation for information
> about how best to update such programs.
>
> On Windows, Command and LookPath now respect the NoDefaultCurrentDirectoryInExePath
> environment variable, making it possible to disable the default implicit search
> of “.” in PATH lookups on Windows systems.
A result of this change was that registering the daemon as a Windows service
no longer worked when done from within the directory of the binary itself:
C:\> cd "Program Files\Docker\Docker\resources"
C:\Program Files\Docker\Docker\resources> dockerd --register-service
exec: "dockerd": cannot run executable found relative to current directory
Note that using an absolute path would work around the issue:
C:\Program Files\Docker\Docker>resources\dockerd.exe --register-service
This patch changes `registerService()` to use `os.Executable()`, instead of
depending on `os.Args[0]` and `exec.LookPath()` for resolving the absolute
path of the binary.
Signed-off-by: Sebastiaan van Stijn <github@gone.nl>
This option is only used for the default bridge network; let's move the
field to that struct to make it clearer what it's used for.
Signed-off-by: Sebastiaan van Stijn <github@gone.nl>
The --mtu option is only used for the default "bridge" network on Linux.
On Windows, the flag is available, but ignored. As this option has been
available for a long time, and was always silently ignored, deprecating
or removing it would be a breaking change (and perhaps it's possible to
support it in future).
This patch:
- hides the option on Windows binaries
- logs a warning if the option is set to any non-zero value other than
the default on a Windows binary
Signed-off-by: Sebastiaan van Stijn <github@gone.nl>
Update docker to support a '--log-format' option, which accepts either
'text' (default) or 'json'. Propagate the log format to containerd as
well, to ensure that everything will be logged consistently.
Signed-off-by: Philip K. Warren <pkwarren@gmail.com>
The existing runtimes reload logic went to great lengths to replace the
directory containing runtime wrapper scripts as atomically as possible
within the limitations of the Linux filesystem ABI. Trouble is,
atomically swapping the wrapper scripts directory solves the wrong
problem! The runtime configuration is "locked in" when a container is
started, including the path to the runC binary. If a container is
started with a runtime which requires a daemon-managed wrapper script
and then the daemon is reloaded with a config which no longer requires
the wrapper script (i.e. some args -> no args, or the runtime is dropped
from the config), that container would become unmanageable. Any attempts
to stop, exec or otherwise perform lifecycle management operations on
the container are likely to fail due to the wrapper script no longer
existing at its original path.
Atomically swapping the wrapper scripts is also incompatible with the
read-copy-update paradigm for reloading configuration. A handler in the
daemon could retain a reference to the pre-reload configuration for an
indeterminate amount of time after the daemon configuration has been
reloaded and updated. It is possible for the daemon to attempt to start
a container using a deleted wrapper script if a request to run a
container races a reload.
Solve the problem of deleting referenced wrapper scripts by ensuring
that all wrapper scripts are *immutable* for the lifetime of the daemon
process. Any given runtime wrapper script must always exist with the
same contents, no matter how many times the daemon config is reloaded,
or what changes are made to the config. This is accomplished by using
everyone's favourite design pattern: content-addressable storage. Each
wrapper script file name is suffixed with the SHA-256 digest of its
contents to (probabilistically) guarantee immutability without needing
any concurrency control. Stale runtime wrapper scripts are only cleaned
up on the next daemon restart.
Split the derived runtimes configuration from the user-supplied
configuration to have a place to store derived state without mutating
the user-supplied configuration or exposing daemon internals in API
struct types. Hold the derived state and the user-supplied configuration
in a single struct value so that they can be updated as an atomic unit.
Signed-off-by: Cory Snider <csnider@mirantis.com>
Ensure data-race-free access to the daemon configuration without
locking by mutating a deep copy of the config and atomically storing
a pointer to the copy into the daemon-wide configStore value. Any
operations which need to read from the daemon config must capture the
configStore value only once and pass it around to guarantee a consistent
view of the config.
Signed-off-by: Cory Snider <csnider@mirantis.com>
Config reloading has interleaved validations and other fallible
operations with mutating the live daemon configuration. The daemon
configuration could be left in a partially-reloaded state if any of the
operations returns an error. Mutating a copy of the configuration and
atomically swapping the config struct on success is not currently an
option as config values are not copyable due to the presence of
sync.Mutex fields. Introduce a two-phase commit protocol to defer any
mutations of the daemon state until after all fallible operations have
succeeded.
Reload transactions are not yet entirely hermetic. The platform
reloading logic for custom runtimes on *nix could still leave the
directory of generated runtime wrapper scripts in an indeterminate state
if an error is encountered.
Signed-off-by: Cory Snider <csnider@mirantis.com>
Historically, daemon.RegistryHosts() has returned a docker.RegistryHosts
callback function which closes over a point-in-time snapshot of the
daemon configuration. When constructing the BuildKit builder at daemon
startup, the return value of daemon.RegistryHosts() has been used.
Therefore the BuildKit builder would use the registry configuration as
it was at daemon startup for the life of the process, even if the
registry configuration is changed and the configuration reloaded.
Provide BuildKit with a RegistryHosts callback which reflects the
live daemon configuration after reloads so that registry operations
performed by BuildKit always use the same configuration as the rest of
the daemon.
Signed-off-by: Cory Snider <csnider@mirantis.com>
Passing around a bare pointer to the map of configured features in order
to propagate to consumers changes to the configuration across reloads is
dangerous. Map operations are not atomic, so concurrently reading from
the map while it is being updated is a data race as there is no
synchronization. Use a getter function to retrieve the current features
map so the features can be retrieved race-free.
Remove the unused features argument from the build router.
Signed-off-by: Cory Snider <csnider@mirantis.com>
These changes add basic CDI integration to the docker daemon.
A cdi driver is added to handle cdi device requests. This
is gated by an experimental feature flag and is only supported on linux
This change also adds a CDISpecDirs (cdi-spec-dirs) option to the config.
This allows the default values of `/etc/cdi`, /var/run/cdi` to be overridden
which is useful for testing.
Signed-off-by: Evan Lezar <elezar@nvidia.com>
This option was deprecated in 5a922dc162, which
is part of the v24.0.0 release, so we can remove it from master.
This patch;
- adds a check to ValidatePlatformConfig, and produces a fatal error
if oom-score-adjust is set
- removes the deprecated libcontainerd/supervisor.WithOOMScore
- removes the warning from docker info
With this patch:
dockerd --oom-score-adjust=-500 --validate
Flag --oom-score-adjust has been deprecated, and will be removed in the next release.
unable to configure the Docker daemon with file /etc/docker/daemon.json: merged configuration validation from file and command line flags failed: DEPRECATED: The "oom-score-adjust" config parameter and the dockerd "--oom-score-adjust" options have been removed.
And when using `daemon.json`:
dockerd --validate
unable to configure the Docker daemon with file /etc/docker/daemon.json: merged configuration validation from file and command line flags failed: DEPRECATED: The "oom-score-adjust" config parameter and the dockerd "--oom-score-adjust" options have been removed.
Signed-off-by: Sebastiaan van Stijn <github@gone.nl>
As of Go 1.8, "net/http".Server provides facilities to close all
listeners, making the same facilities in server.Server redundant.
http.Server also improves upon server.Server by additionally providing a
facility to also wait for outstanding requests to complete after closing
all listeners. Leverage those facilities to give in-flight requests up
to five seconds to finish up after all containers have been shut down.
Signed-off-by: Cory Snider <csnider@mirantis.com>
Logging through a dependency-injected interface value was a vestige of
when Trap was in pkg/signal to avoid importing logrus in a reusable
package: cc4da81128.
Now that Trap lives under cmd/dockerd, nobody will be importing this so
we no longer need to worry about minimizing the package's dependencies.
Signed-off-by: Cory Snider <csnider@mirantis.com>
Always calling os.Exit() on clean shutdown may not always be desirable
as deferred functions are not run. Let the cleanup callback decide
whether or not to call os.Exit() itself. Allow the process to exit the
normal way, by returning from func main().
Simplify the trap.Trap implementation. The signal notifications are
buffered in a channel so there is little need to spawn a new goroutine
for each received signal. With all signals being handled in the same
goroutine, there are no longer any concurrency concerns around the
interrupt counter.
Signed-off-by: Cory Snider <csnider@mirantis.com>
The image store sends events when a new image is created/tagged, using
it instead of the reference store makes sure we send the "tag" event
when a new image is built using buildx.
Signed-off-by: Djordje Lukic <djordje.lukic@docker.com>
The fix to ignore SIGPIPE signals was originally added in the Go 1.4
era. signal.Ignore was first added in Go 1.5.
Signed-off-by: Cory Snider <csnider@mirantis.com>
