As pointed out in #1459, docker cli fails to detect that the input is a tarball,
in case it is generated by `git archive --format=tgz`.
This happens because `git archive` adds some metadata to the initial tar header,
and so it is more than 1 block (of 512 bytes) long, while we only provide 1 block
to archive/tar.Next() and it fails.
To fix, give it 2 blocks :)
Signed-off-by: Kir Kolyshkin <kolyshkin@gmail.com>
This adds validation to `docker container run` / `docker container create`;
Validation of labels provided through flags was removed in 31dc5c0a9a,
after the validation was changed to fix labels without values, and to prevent
labels from being expanded with environment variables in 2b17f4c8a8
However, now empty label names from _files_ (`--label-file`) followed different
validation rules than labels passed through `--label`.
This patch adds back minimal validation for labels passed through the command-line
Before this patch:
```bash
docker container create \
--name label \
--label==with-leading-equal-sign \
--label=without-value \
--label=somelabel=somevalue \
--label " = " \
--label=with-quotes-in-value='{"foo"}' \
--label='with"quotes"in-key=test' \
busybox
docker container inspect --format '{{json .Config.Labels}}' label
```
```json
{
"": "with-leading-equal-sign",
" ": " ",
"somelabel": "somevalue",
"with\"quotes\"in-key": "test",
"with-quotes-in-value": "{\"foo\"}",
"without-value": ""
}
```
After this patch:
```bash
docker container create \
--name label \
--label==with-leading-equal-sign \
--label=without-value \
--label=somelabel=somevalue \
--label " = " \
--label=with-quotes-in-value='{"foo"}' \
--label='with"quotes"in-key=test' \
busybox
invalid argument "=with-leading-equal-sign" for "-l, --label" flag: invalid label format: "=with-leading-equal-sign"
```
Signed-off-by: Sebastiaan van Stijn <github@gone.nl>
This patch fixes a bug where labels use the same behavior as `--env`, resulting
in a value to be copied from environment variables with the same name as the
label if no value is set (i.e. a simple key, no `=` sign, no value).
An earlier pull request addressed similar cases for `docker run`;
2b17f4c8a8, but this did not address the
same situation for (e.g.) `docker service create`.
Digging in history for this bug, I found that use of the `ValidateEnv`
function for labels was added in the original implementation of the labels feature in
abb5e9a077 (diff-ae476143d40e21ac0918630f7365ed3cR34)
However, the design never intended it to expand environment variables,
and use of this function was either due to either a "copy/paste" of the
equivalent `--env` flags, or a misunderstanding (the name `ValidateEnv` does
not communicate that it also expands environment variables), and the existing
`ValidateLabel` was designed for _engine_ labels (which required a value to
be set).
Following the initial implementation, other parts of the code followed
the same (incorrect) approach, therefore leading the bug to be introduced
in services as well.
This patch:
- updates the `ValidateLabel` to match the expected validation
rules (this function is no longer used since 31dc5c0a9a),
and the daemon has its own implementation)
- corrects various locations in the code where `ValidateEnv` was used instead of `ValidateLabel`.
Before this patch:
```bash
export SOME_ENV_VAR=I_AM_SOME_ENV_VAR
docker service create --label SOME_ENV_VAR --tty --name test busybox
docker service inspect --format '{{json .Spec.Labels}}' test
{"SOME_ENV_VAR":"I_AM_SOME_ENV_VAR"}
```
After this patch:
```bash
export SOME_ENV_VAR=I_AM_SOME_ENV_VAR
docker service create --label SOME_ENV_VAR --tty --name test busybox
docker container inspect --format '{{json .Config.Labels}}' test
{"SOME_ENV_VAR":""}
```
Signed-off-by: Sebastiaan van Stijn <github@gone.nl>
The warning, printed before running `docker system prune` was printing the
filters in JSON format.
This patch attempts to make the output human readable;
- updating the code, and template to print filters individually
- reducing the indentation (which was quite deep)
Before this patch was applied;
```
docker system prune --filter until=24h --filter label=hello-world --filter label!=foo=bar --filter label=bar=baz
WARNING! This will remove:
- all stopped containers
- all networks not used by at least one container
- all dangling images
- all dangling build cache
- Elements to be pruned will be filtered with:
- label={"label":{"bar=baz":true,"hello-world":true},"label!":{"foo=bar":true},"until":{"24h":true}}
Are you sure you want to continue? [y/N]
```
With this patch applied;
```
WARNING! This will remove:
- all stopped containers
- all networks not used by at least one container
- all dangling images
- all dangling build cache
Items to be pruned will be filtered with:
- label!=foo=bar
- label!=never=remove-me
- label=bar=baz
- label=hello-world
- label=remove=me
- until=24h
Are you sure you want to continue? [y/N]
```
Signed-off-by: Sebastiaan van Stijn <github@gone.nl>
The warning, printed before runing docker system prune was missing any filter
that was set in the configuration file. In addition, the warning prefixes the
filters with `label=`, which is no longer accurate, now that the prune command
also supports "until" as a filter.
Before this change, only the filters set on the command-line were shown,
and any filter set in the configuration file was missing;
```
mkdir -p ./test-config
echo '{"pruneFilters": ["label!=never=remove-me", "label=remove=me"]}' > test-config/config.json
docker --config=./test-config system prune --filter until=24h --filter label=hello-world --filter label!=foo=bar --filter label=bar=baz
WARNING! This will remove:
- all stopped containers
- all networks not used by at least one container
- all dangling images
- all dangling build cache
- Elements to be pruned will be filtered with:
- label={"label":{"bar=baz":true,"hello-world":true},"label!":{"foo=bar":true},"until":{"24h":true}}
Are you sure you want to continue? [y/N]
```
With this patch applied, both options from the commandline and options set
in the configuration file are shown;
```
mkdir -p ./test-config
echo '{"pruneFilters": ["label!=never=remove-me", "label=remove=me"]}' > test-config/config.json
docker --config=./test-config system prune --filter until=24h --filter label=hello-world --filter label!=foo=bar --filter label=bar=baz
WARNING! This will remove:
- all stopped containers
- all networks not used by at least one container
- all dangling images
- all dangling build cache
- Elements to be pruned will be filtered with:
- filter={"label":{"bar=baz":true,"hello-world":true,"remove=me":true},"label!":{"foo=bar":true,"never=remove-me":true},"until":{"24h":true}}
```
Signed-off-by: Sebastiaan van Stijn <github@gone.nl>
- when using "--context default" parameter
- when printing the list of contexts
- when exporting the default context to a tarball
Signed-off-by: Jean-Christophe Sirot <jean-christophe.sirot@docker.com>
(+1 squashed commit)
Squashed commits:
[20670495] Fix CLI initialization for the `docker stack deploy --help` command and ensure that the dockerCli.CurrentContext() always returns a non empty context name (default as a fallback)
Remove now obsolete code handling empty string context name
Minor code cleanup
Signed-off-by: Jean-Christophe Sirot <jean-christophe.sirot@docker.com>
This happens on Windows when dialing a named pipe (a path which is used by CLI
plugins), in that case some debugging shows:
DEBU[0000] conn is a *winio.win32MessageBytePipe
DEBU[0000] conn is a halfReadCloser: false
DEBU[0000] conn is a halfWriteCloser: true
the raw stream connection does not implement halfCloser
In such cases we can simply wrap with a nop function since closing for read
isn't too critical.
Signed-off-by: Ian Campbell <ijc@docker.com>
The `conn` here is `*winio.win32MessageBytePipe` which does not have a
`CloseRead` method (it does have `CloseWrite`) resulting in:
docker@WIN-NUC0 C:\Users\docker>.\docker-windows-amd64.exe system dial-stdio
the raw stream connection does not implement halfCloser
Also disable the path which uses this for cli-plugins on Windows.
Signed-off-by: Ian Campbell <ijc@docker.com>
This means that plugins can use whatever methods the monolithic CLI supports,
which is good for consistency.
This relies on `os.Args[0]` being something which can be executed again to
reach the same binary, since it is propagated (via an envvar) to the plugin for
this purpose. This essentially requires that the current working directory and
path are not modified by the monolithic CLI before it launches the plugin nor
by the plugin before it initializes the client. This should be the case.
Previously the fake apiclient used by `TestExperimentalCLI` was not being used,
since `cli.Initialize` was unconditionally overwriting it with a real one
(talking to a real daemon during unit testing, it seems). This wasn't expected
nor desirable and no longer happens with the new arrangements, exposing the
fact that no `pingFunc` is provided, leading to a panic. Add a `pingFunc` to
the fake client to avoid this.
Signed-off-by: Ian Campbell <ijc@docker.com>
Fairly straight forward. It became necessary to wrap `Plugin.Err` with a type
which implements `encoding.MarshalText` in order to have that field rendered
properly in the `docker info -f '{{json}}'` output.
Since I changed the type somewhat I also added a unit test for `formatInfo`.
Signed-off-by: Ian Campbell <ijc@docker.com>
- The `/build/prune` endpoint was added in API v1.31
- The `/network` endpoints were added in API v1.21
This patch hides these commands on older API versions
Before this change:
```
DOCKER_API_VERSION=1.0 docker
...
Management Commands:
builder Manage builds
container Manage containers
image Manage images
manifest Manage Docker image manifests and manifest lists
network Manage networks
system Manage Docker
trust Manage trust on Docker images
```
After this change
```
DOCKER_API_VERSION=1.0 docker
...
Management Commands:
container Manage containers
image Manage images
manifest Manage Docker image manifests and manifest lists
system Manage Docker
trust Manage trust on Docker images
```
Signed-off-by: Sebastiaan van Stijn <github@gone.nl>
I authored this for `contentTrustEnabled` prior to 7f207f3f95, so this now
tests the funcation argument version.
Signed-off-by: Ian Campbell <ijc@docker.com>
This helps to avoid circular includes, by separating the pure data out from the
actual functionality in the cli subpackage, allowing other code which is
imported to access the data.
Signed-off-by: Ian Campbell <ijc@docker.com>
All of the current callers follow the pattern:
dockerPreRun(opts)
err := dockerCli.Initialize(opts) ...
So there is no semantic change into merging the content of `dockerPreRun` into the head of `Initialize`.
I'm about to add a new caller outside of the `cmd/docker` package and this
seems preferable exporting `DockerPreRun`.
Signed-off-by: Ian Campbell <ijc@docker.com>
This will allow plugins to have custom typed endpoints, as well as
create/remove/update contexts with the exact same results as the main
CLI (thinking of things like `docker ee login https://my-ucp-server
--context ucp-prod)`
Signed-off-by: Simon Ferquel <simon.ferquel@docker.com>
That is, reindent the two sections by one space.
While the code was done by hand the `.golden` files had the extra space
inserted with emacs' `string-insert-rectangle` macro to (try to) avoid possible
manual errors. The docs were edited the same way.
Signed-off-by: Ian Campbell <ijc@docker.com>
Right now the only client side info we have is whether debug is enabled, but we
expect more in the future.
We also preemptively prepare for the possibility of multiple errors when
gathering both daemon and client info.
Signed-off-by: Ian Campbell <ijc@docker.com>
Proxies configured in config.json were only taking effect
when using `docker run`, but were being ignored when
using `docker create`.
Before this change:
echo '{"proxies":{"default":{"httpProxy":"httpProxy","httpsProxy":"httpsProxy","noProxy":"noProxy","ftpProxy":"ftpProxy"}}}' > config.json
docker inspect --format '{{.Config.Env}}' $(docker --config=./ create busybox)
[PATH=/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin]
With this change applied:
echo '{"proxies":{"default":{"httpProxy":"httpProxy","httpsProxy":"httpsProxy","noProxy":"noProxy","ftpProxy":"ftpProxy"}}}' > config.json
docker inspect --format '{{.Config.Env}}' $(docker --config=./ create busybox)
[NO_PROXY=noProxy no_proxy=noProxy FTP_PROXY=ftpProxy ftp_proxy=ftpProxy HTTP_PROXY=httpProxy http_proxy=httpProxy HTTPS_PROXY=httpsProxy https_proxy=httpsProxy PATH=/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin]
Reported-by: Silvano Cirujano Cuesta <Silvanoc@users.noreply.github.com>
Signed-off-by: Sebastiaan van Stijn <github@gone.nl>
The `DOCKER_HIDE_LEGACY_COMMANDS` environment variable allows hiding legacy
top-level commands that are now available under `docker <object> <verb>`. The
`docker deploy` top-level command is experimental, and replaced by
`docker stack deploy`.
This patch hides the top-level `docker deploy` if the `DOCKER_HIDE_LEGACY_COMMANDS`
environment variable is set.
Before this change:
DOCKER_HIDE_LEGACY_COMMANDS=1 docker --help
...
Commands:
build Build an image from a Dockerfile
deploy Deploy a new stack or update an existing stack
login Log in to a Docker registry
logout Log out from a Docker registry
run Run a command in a new container
search Search the Docker Hub for images
version Show the Docker version information
...
With this patch applied:
DOCKER_HIDE_LEGACY_COMMANDS=1 docker --help
...
Commands:
build Build an image from a Dockerfile
login Log in to a Docker registry
logout Log out from a Docker registry
run Run a command in a new container
search Search the Docker Hub for images
version Show the Docker version information
...
Signed-off-by: Sebastiaan van Stijn <github@gone.nl>
This PR adds a store to the CLI, that can be leveraged to persist and
retrieve credentials for various API endpoints, as well as
context-specific settings (initially, default stack orchestrator, but we
could expand that).
This comes with the logic to persist and retrieve endpoints configs
for both Docker and Kubernetes APIs.
Signed-off-by: Simon Ferquel <simon.ferquel@docker.com>
Add `--quiet` to the `docker image pull` subcommand that will not pull
the image quietly.
```
$ docker pull -q golang
Using default tag: latest
```
Signed-off-by: Vincent Demeester <vincent@sbr.pm>
Previously, these errors were only printed when using `docker run`, but were
omitted when using `docker container create` and `docker container start`
separately.
Given that these warnings apply to both situations, this patch moves generation
of these warnings to `docker container create` (which is also called by
`docker run`)
Signed-off-by: Sebastiaan van Stijn <github@gone.nl>
A while ago, Docker split the "Domainname" field out from the "Hostname"
field for the container configuration. There was no real user-visible
change associated with this (and under the hood "Domainname" was mostly
left unused from the command-line point of view). We now add this flag
in order to match other proposed changes to allow for setting the NIS
domainname of a container.
This also includes a fix for the --hostname parsing tests (they would
not error out if only one of .Hostname and .Domainname were incorrectly
set -- which is not correct).
Signed-off-by: Aleksa Sarai <asarai@suse.de>
This PR chnages allow user to configure data path
port number. By default we use 4789 port number. But this commit
will allow user to configure port number during swarm init.
Data path port can't be modified after swarm init.
Signed-off-by: selansen <elango.siva@docker.com>
A recent change in moby/moby made tests with missing client mocks fail with panic.
This adds those missing mocks for the impacted tests.
Signed-off-by: Simon Ferquel <simon.ferquel@docker.com>
- make it possible to extract the formatter implementation from the
"common" code, that way, the formatter package stays small
- extract some formatter into their own packages
This is essentially moving the "formatter" implementation of each type
in their respective packages. The *main* reason to do that, is to be
able to depend on `cli/command/formatter` without depending of the
implementation detail of the formatter. As of now, depending on
`cli/command/formatter` means we depend on `docker/docker/api/types`,
`docker/licensing`, … — that should not be the case. `formatter`
should hold the common code (or helpers) to easily create formatter,
not all formatter implementations.
Signed-off-by: Vincent Demeester <vincent@sbr.pm>
Avoid testing for session support in non-buildkit builder to support
servers that falsely report as `1.39` compatible
Signed-off-by: Tibor Vass <tibor@docker.com>
gofmt/goimports changed some heuristics in 1.11 and the code is now
formatted slightly differently.
No functional change, just whitespace.
Signed-off-by: Kir Kolyshkin <kolyshkin@gmail.com>
Unlike `docker build --secret`, `docker build --ssh` allows the build container to
use SSH keys with passphrases.
$ eval $(ssh-agent)
$ ssh-add ~/.ssh/id_rsa
(Input your passphrase here)
$ docker build --ssh default=$SSH_AUTH_SOCK ...
This feature requires the daemon with `CapExecMountSSH` build capability (moby/moby#37973) .
Currently, the official Dockerfile frontend does not provide the syntax for using the SSH forwarder.
However, the experimental `RUN --mount=type=ssh` syntax can be enabled by using
the Dockerfile frontend image built with the `BUILDTAGS="dfrunmount dfssh"`, via the `# syntax =` "shebang".
The Dockerfile for the Dockerfile frontend is available at github.com/moby/buildkit/frontend/dockerfile/cmd/dockerfile-frontend)
The pre-built image is also available as `tonistiigi/dockerfile:ssh20181002` .
An example Dockerfile with `RUN --mount=type=ssh`:
# syntax = tonistiigi/dockerfile:ssh20181002
FROM alpine
RUN apk add --no-cache openssh-client
RUN mkdir -p -m 0700 ~/.ssh && ssh-keyscan gitlab.com >> ~/.ssh/known_hosts
RUN --mount=type=ssh ssh git@gitlab.com | tee /hello
# "Welcome to GitLab, @GITLAB_USERNAME_ASSOCIATED_WITH_SSHKEY" should be printed here
More info available at moby/buildkit#608, moby/buildkit#655
Signed-off-by: Akihiro Suda <suda.akihiro@lab.ntt.co.jp>
Signed-off-by: Daniel Hiltgen <daniel.hiltgen@docker.com>
(cherry picked from commit 92932647d3)
Signed-off-by: Daniel Hiltgen <daniel.hiltgen@docker.com>
Prior refactoring passes missed a corner case.
Signed-off-by: Daniel Hiltgen <daniel.hiltgen@docker.com>
(cherry picked from commit dee37936e5)
Signed-off-by: Daniel Hiltgen <daniel.hiltgen@docker.com>
During the refactoring for 18.09 the activate/update flows no longer
restart the engine explicitly but let the user do that when they're ready,
so the health check logic is no longer required.
Signed-off-by: Daniel Hiltgen <daniel.hiltgen@docker.com>
(cherry picked from commit f2b2061cc3)
Signed-off-by: Sebastiaan van Stijn <github@gone.nl>
This is a follow up PR to #1381 to address some of the review comments
we didn't get to.
Signed-off-by: Daniel Hiltgen <daniel.hiltgen@docker.com>
(cherry picked from commit c12e23a4c1)
Signed-off-by: Daniel Hiltgen <daniel.hiltgen@docker.com>
Help the user understand which license they're about
to load in case they have multiple licenses they need to
figure out.
Signed-off-by: Daniel Hiltgen <daniel.hiltgen@docker.com>
(cherry picked from commit 5a97a93ae1)
Signed-off-by: Daniel Hiltgen <daniel.hiltgen@docker.com>
Signed-off-by: Daniel Hiltgen <daniel.hiltgen@docker.com>
(cherry picked from commit 6004d74b1f)
Signed-off-by: Daniel Hiltgen <daniel.hiltgen@docker.com>
Address code review comemnts and purge additional dead code.
Signed-off-by: Daniel Hiltgen <daniel.hiltgen@docker.com>
(cherry picked from commit f250152bf4)
Signed-off-by: Daniel Hiltgen <daniel.hiltgen@docker.com>
Adapt the CLI to the host install model for 18.09.
Signed-off-by: Daniel Hiltgen <daniel.hiltgen@docker.com>
(cherry picked from commit 342afe44fb)
Signed-off-by: Daniel Hiltgen <daniel.hiltgen@docker.com>
Signed-off-by: Michael Crosby <crosbymichael@gmail.com>
(cherry picked from commit cfec8027ed)
Signed-off-by: Daniel Hiltgen <daniel.hiltgen@docker.com>
The integration test TestExportContainerWithOutputAndImportImage in moby/moby is the same as TestExportContainerAndImportImage,
except for the output file option. Adding a unit test to cover the output file option of the export command here allows
the removal of the redundant integration test TestExportContainerWithOutputAndImportImage.
Signed-off-by: Arash Deshmeh <adeshmeh@ca.ibm.com>
Signed-off-by: Daniel Hiltgen <daniel.hiltgen@docker.com>
(cherry picked from commit 68be7cb376)
Signed-off-by: Daniel Hiltgen <daniel.hiltgen@docker.com>
Warnings are now generated by the daemon, and returned as
part of the /info API response.
If warnings are returned by the daemon; use those instead
of generating them locally.
Signed-off-by: Sebastiaan van Stijn <github@gone.nl>
This feature brings new attribute/option for swarm init command.
default-addr-pool will take string input which can be in below format.
"CIDR,CIDR,CIDR...:SUBNET-SIZE".
Signed-off-by: selansen <elango.siva@docker.com>
This new collection of commands supports initializing a local
engine using containerd, updating that engine, and activating
the EE product
Signed-off-by: Daniel Hiltgen <daniel.hiltgen@docker.com>
This patch adds a new builder subcommand, allowing to add more builder-related
commands in the future. Unfortunately `build` expects an argument so could not
be used as a subcommand.
This also implements `docker builder prune`, which is needed to prune the builder
cache manually without having to call `docker system prune`.
Today when relying on the legacy builder, users are able to prune dangling images
(used as build cache) by running `docker image prune`. This patch allows the
same usecase with buildkit.
Signed-off-by: Tibor Vass <tibor@docker.com>
This patch implements `docker build --secret id=mysecret,src=/secret/file`
for buildkit frontends that request the mysecret secret.
It is currently implemented in the tonistiigi/dockerfile:secrets20180808
frontend via RUN --mount=type=secret,id=mysecret
Signed-off-by: Tibor Vass <tibor@docker.com>
This changes the experimental --console flag to --progress following
feedback indicating avoidable confusion.
In addition to naming changes, the help output now has an additional
clarification, specifically: container output during builds are only
shown when progress output is set to plain. Not mentioning this was also
a big cause of confusion.
Signed-off-by: Tibor Vass <tibor@docker.com>
e.g. docker -H ssh://me@server
The `docker` CLI also needs to be installed on the remote host to
provide `docker system dial-stdio`, which proxies the daemon socket to stdio.
Please refer to docs/reference/commandline/dockerd.md .
Signed-off-by: Akihiro Suda <suda.akihiro@lab.ntt.co.jp>
PersistentPreRunE needs to be called within the help function to initialize all the flags (notably the orchestrator flag)
Add an e2e test as regression test
Signed-off-by: Silvin Lubecki <silvin.lubecki@docker.com>
This flag was added in Docker 17.06, API version 1.31 through
moby@8dc8cd4719f165c01c98e7d3ce1d6cea6a8f60b8, but didn't add
API-version annotations.
This patch adds the missing annotations to hide this flag if
the CLI is connected to an older version of the daemon that
doesn't support that API.
Before this patch:
DOCKER_API_VERSION=1.30 docker swarm init --help | grep data-path-addr
--data-path-addr string Address or interface to use for data path traffic (format: <ip|interface>)
DOCKER_API_VERSION=1.31 docker swarm init --help | grep data-path-addr
--data-path-addr string Address or interface to use for data path traffic (format: <ip|interface>)
With this patch applied:
DOCKER_API_VERSION=1.30 docker swarm init --help | grep data-path-addr
# (no result)
DOCKER_API_VERSION=1.31 docker swarm init --help | grep data-path-addr
--data-path-addr string Address or interface to use for data path traffic (format: <ip|interface>)
Signed-off-by: Sebastiaan van Stijn <github@gone.nl>
- Use `Contains` instead of `Include`
- Use `ToJSON` instead of `ToParam`
- Remove usage of `ParseFlag` as it is deprecated too
Signed-off-by: Vincent Demeester <vincent@sbr.pm>
Clarified ambiguous error message
Update kubernetes/cli.go
Infromed user of why the error was caused when file is not there
Signed-off-by: Justyn Temme <justyntemme@gmail.com>
- remove some hints that are no longer needed
- added a nolint: unparam for removeSingleSigner() (return bool is only used in tests)
Signed-off-by: Sebastiaan van Stijn <github@gone.nl>
in swarm.
Also, fix some CLI command confusions:
1. If the --external-ca flag is provided, require a --ca-cert flag as well, otherwise
the external CA is set but the CA certificate is actually rotated to an internal
cert
2. If a --ca-cert flag is provided, require a --ca-key or --external-ca flag be
provided as well, otherwise either the server will say that the request is
invalid, or if there was previously an external CA corresponding to the cert, it
will succeed. While that works, it's better to require the user to explicitly
set all the parameters of the new desired root CA.
This also changes the `swarm update` function to set the external CA's CACert field,
which while not strictly necessary, makes the CA list more explicit.
Signed-off-by: Ying Li <ying.li@docker.com>
Stores complete OCI descriptor instead of digest and platform
fields. This includes the size which was getting lost by not
storing the original manifest bytes.
Attempt to support existing cached files, if not output
the filename with the incorrect content.
Signed-off-by: Derek McGowan <derek@mcgstyle.net>
It makes it easier to get the correct stack from a compose config
struct without requiring the client (and thus talking to k8s API)
Signed-off-by: Vincent Demeester <vincent@sbr.pm>
- Handle `bundlefile` directly in the `top-level`
command. `bundlefile` is still experimental and will be deprecated
in future version — this should make be easier to remove it.
- Validate the `stack` name in all cases (i.e. whatever the
orchestrator is used)
- Load the composefile ahead of choosing the orchestrator. This
removes some slight duplication.
- Makes `RunDeploy` easier to use from outside packages (like
`docker/app`) with a preloaded configuration.
Signed-off-by: Vincent Demeester <vincent@sbr.pm>
The output of this information can be confusing,
so removing until we have a better design for this.
Signed-off-by: Sebastiaan van Stijn <github@gone.nl>
* Renaming DOCKER_ORCHESTRATOR to DOCKER_STACK_ORCHESTRATOR
* Renaming config file option "orchestrator" to "stackOrchestrator"
* "--orchestrator" flag is no more global but local to stack command and subcommands
* Cleaning all global orchestrator code
* Replicating Hidden flags in help and Supported flags from root command to stack command
Signed-off-by: Silvin Lubecki <silvin.lubecki@docker.com>
Unfortunately, this is for now the only way to see the output of RUN commands when using buildkit.
It is equivalent to `DOCKER_BUILDKIT=1 docker build . 2>&1 | cat`
Signed-off-by: Tibor Vass <tibor@docker.com>
With this patch the following become true even with buildkit enabled:
1. `docker build -q .` only outputs the created image's sha256 ID.
2. `docker build -q .` outputs as if no `-q` was specified, if error occurred
3. `docker build . &> out` outputs JSON (instead of TTY characters)
Signed-off-by: Tibor Vass <tibor@docker.com>
This commit brings a more pedantic change in the following ambiguous case:
cat Dockerfile | docker build -f otherDockerfile -
The legacy builder does not error out and prefers the Dockerfile from stdin
while the buildkit-based one errors out.
Note that this is only in the case where stdin is a Dockerfile (not an archive)
Signed-off-by: Tibor Vass <tibor@docker.com>
* Kubernetes native filtering (server side) is an exact match, now filtering on name is made client-side to add prefix-matching
Signed-off-by: Silvin Lubecki <silvin.lubecki@docker.com>
This patch adds annotations to mark the checkpoint commands as Linux only, which
hides them if the daemon is running a non-matching operating-system type;
Before:
docker
Usage: docker COMMAND
A self-sufficient runtime for containers
...
Management Commands:
config Manage Docker configs
container Manage containers
image Manage images
After:
docker
Usage: docker COMMAND
A self-sufficient runtime for containers
...
Management Commands:
checkpoint Manage checkpoints
config Manage Docker configs
container Manage containers
image Manage images
This change also prints errors when attempting to use checkpoint commands or
flags if the feature is not supported by the Daemon's operating system;
$ docker checkpoint --help
docker checkpoint is only supported on a Docker daemon running on linux, but the Docker daemon is running on windows
$ docker checkpoint create --help
docker checkpoint create is only supported on a Docker daemon running on linux, but the Docker daemon is running on windows
$ docker checkpoint ls --help
docker checkpoint ls is only supported on a Docker daemon running on linux, but the Docker daemon is running on windows
$ docker checkpoint rm --help
docker checkpoint rm is only supported on a Docker daemon running on linux, but the Docker daemon is running on windows
$ docker container start --checkpoint=foo mycontainer
"--checkpoint" requires the Docker daemon to run on linux, but the Docker daemon is running on windows
$ docker container start --checkpoint-dir=/foo/bar mycontainer
"--checkpoint-dir" requires the Docker daemon to run on linux, but the Docker daemon is running on windows
Signed-off-by: Sebastiaan van Stijn <github@gone.nl>
Add validation for stack names to prevent an empty name resulting in _all_
stacks to be returned after filtering, which can result in removal of services
for all stacks if `--prune`, or `docker stack rm` is used.
Before this change;
docker stack deploy -c docker-compose.yml one
docker stack deploy -c docker-compose.yml two
docker stack deploy -c docker-compose.yml three
docker stack deploy -c docker-compose.yml --prune ''
Removing service one_web
Removing service two_web
Removing service three_web
After this change:
docker stack deploy -c docker-compose.yml one
docker stack deploy -c docker-compose.yml two
docker stack deploy -c docker-compose.yml three
docker stack deploy -c docker-compose.yml --prune ''
invalid stack name: ""
Other stack commands were updated as well:
Before this change;
docker stack deploy -c docker-compose.yml ''
Creating network _default
failed to create network _default: Error response from daemon: rpc error: code = InvalidArgument desc = name must be valid as a DNS name component
docker stack ps ''
nothing found in stack:
docker stack rm ''
Removing service one_web
Removing service three_web
Removing service two_web
After this change:
docker stack deploy -c docker-compose.yml ''
invalid stack name: ""
docker stack ps ''
invalid stack name: ""
docker stack rm ''
invalid stack name: ""
Signed-off-by: Sebastiaan van Stijn <github@gone.nl>
- Don't wrap the code to be slightly more readable
- Rename `getServiceFilter()` to `getStackServiceFilter()` to be
consistent with other helper functions.
Signed-off-by: Sebastiaan van Stijn <github@gone.nl>
* Add "kubernetes" struct in config file with "allNamespaces" option, to opt-out this behavior when set as "disabled"
Signed-off-by: Mathieu Champlon <mathieu.champlon@docker.com>
`docker stack services --filter=label=foo=bar --filter=label=foo=baz my-stack` with Swarm gets handled as `filter on (a label named foo with value bar) AND (a label named foo with value baz).
This obviously yields an empty result set every time, but if and how this should be changed is out of scope here, so simply align Kubernetes with Swarm for now.
Signed-off-by: Mathieu Champlon <mathieu.champlon@docker.com>
Before this change:
----------------------------------------------------
Create a service with reservations and limits for memory and cpu:
docker service create --name test \
--limit-memory=100M --limit-cpu=1 \
--reserve-memory=100M --reserve-cpu=1 \
nginx:alpine
Verify the configuration
docker service inspect --format '{{json .Spec.TaskTemplate.Resources}}' test
{
"Limits": {
"NanoCPUs": 1000000000,
"MemoryBytes": 104857600
},
"Reservations": {
"NanoCPUs": 1000000000,
"MemoryBytes": 104857600
}
}
Update just CPU limit and reservation:
docker service update --limit-cpu=2 --reserve-cpu=2 test
Notice that the memory limit and reservation is not preserved:
docker service inspect --format '{{json .Spec.TaskTemplate.Resources}}' test
{
"Limits": {
"NanoCPUs": 2000000000
},
"Reservations": {
"NanoCPUs": 2000000000
}
}
Update just Memory limit and reservation:
docker service update --limit-memory=200M --reserve-memory=200M test
Notice that the CPU limit and reservation is not preserved:
docker service inspect --format '{{json .Spec.TaskTemplate.Resources}}' test
{
"Limits": {
"MemoryBytes": 209715200
},
"Reservations": {
"MemoryBytes": 209715200
}
}
After this change:
----------------------------------------------------
Create a service with reservations and limits for memory and cpu:
docker service create --name test \
--limit-memory=100M --limit-cpu=1 \
--reserve-memory=100M --reserve-cpu=1 \
nginx:alpine
Verify the configuration
docker service inspect --format '{{json .Spec.TaskTemplate.Resources}}' test
{
"Limits": {
"NanoCPUs": 1000000000,
"MemoryBytes": 104857600
},
"Reservations": {
"NanoCPUs": 1000000000,
"MemoryBytes": 104857600
}
}
Update just CPU limit and reservation:
docker service update --limit-cpu=2 --reserve-cpu=2 test
Confirm that the CPU limits/reservations are updated, but memory limit and reservation are preserved:
docker service inspect --format '{{json .Spec.TaskTemplate.Resources}}' test
{
"Limits": {
"NanoCPUs": 2000000000,
"MemoryBytes": 104857600
},
"Reservations": {
"NanoCPUs": 2000000000,
"MemoryBytes": 104857600
}
}
Update just Memory limit and reservation:
docker service update --limit-memory=200M --reserve-memory=200M test
Confirm that the Mempry limits/reservations are updated, but CPU limit and reservation are preserved:
docker service inspect --format '{{json .Spec.TaskTemplate.Resources}}' test
{
"Limits": {
"NanoCPUs": 2000000000,
"MemoryBytes": 209715200
},
"Reservations": {
"NanoCPUs": 2000000000,
"MemoryBytes": 209715200
}
}
Signed-off-by: Sebastiaan van Stijn <github@gone.nl>
Signed-off-by: Sebastiaan van Stijn <github@gone.nl>
The `docker version` output now uses a tabwriter, so use single
tabs to print the output.
Before this change:
Server:
Engine:
Version: 18.05.0-ce
API version: 1.37 (minimum version 1.12)
Go version: go1.10.1
Git commit: f150324
Built: Wed May 9 22:20:16 2018
OS/Arch: linux/amd64
Experimental: true
Kubernetes:
Version: v1.9.6
StackAPI: v1beta2
After this change:
Server:
Engine:
Version: 18.05.0-ce
API version: 1.37 (minimum version 1.12)
Go version: go1.10.1
Git commit: f150324
Built: Wed May 9 22:20:16 2018
OS/Arch: linux/amd64
Experimental: true
Kubernetes:
Version: v1.9.6
StackAPI: v1beta2
Signed-off-by: Sebastiaan van Stijn <github@gone.nl>
Switch from x/net/context to context made "go vet" see the previously
unseen errors:
> cli/command/container/start.go:57::error: the cancelFun function is
> not used on all paths (possible context leak) (vet)
> cli/command/container/start.go:63::error: this return statement may be
> reached without using the cancelFun var defined on line 57 (vet)
> cli/command/container/run.go:159::error: the cancelFun function is not
> used on all paths (possible context leak) (vet)
> cli/command/container/run.go:164::error: this return statement may be
> reached without using the cancelFun var defined on line 159 (vet)
Do call the cancel function.
Note we might end up calling it twice which is fine as long as I can see
from the Go 1.10 source code.
Signed-off-by: Kir Kolyshkin <kolyshkin@gmail.com>
Since go 1.7, "context" is a standard package. Since go 1.9,
x/net/context merely provides some types aliased to those in
the standard context package.
The changes were performed by the following script:
for f in $(git ls-files \*.go | grep -v ^vendor/); do
sed -i 's|golang.org/x/net/context|context|' $f
goimports -w $f
for i in 1 2; do
awk '/^$/ {e=1; next;}
/\t"context"$/ {e=0;}
{if (e) {print ""; e=0}; print;}' < $f > $f.new && \
mv $f.new $f
goimports -w $f
done
done
[v2: do awk/goimports fixup twice]
Signed-off-by: Kir Kolyshkin <kolyshkin@gmail.com>
Removing a host by `<host>:<ip>` should only remove occurences of the host with
a matching IP-address, instead of removing all entries for that host.
In addition, combining `--host-rm` and `--host-add` for the same host should
result in the new host being added.
This patch fixes the way the diff is calculated to allow combining
removing/adding, and to support entries having both a canonical, and aliases.
Aliases cannot be added by the CLI, but are supported in the Service spec, thus
should be taken into account:
Entries can be removed by either a specific `<host-name>:<ip-address>`
mapping, or by `<host>` alone:
- If both IP-address and host-name is provided, only remove the hostname
from entries that match the given IP-address.
- If only a host-name is provided, remove the hostname from any entry it
is part of (either as _canonical_ host-name, or as _alias_).
- If, after removing the host-name from an entry, no host-names remain in
the entry, the entry itself should be removed.
For example, the list of host-entries before processing could look like this:
hosts = &[]string{
"127.0.0.2 host3 host1 host2 host4",
"127.0.0.1 host1 host4",
"127.0.0.3 host1",
"127.0.0.1 host1",
}
Removing `host1` removes every occurrence:
hosts = &[]string{
"127.0.0.2 host3 host2 host4",
"127.0.0.1 host4",
}
Whereas removing `host1:127.0.0.1` only remove the host if the IP-address matches:
hosts = &[]string{
"127.0.0.2 host3 host1 host2 host4",
"127.0.0.1 host4",
"127.0.0.3 host1",
}
Before this patch:
$ docker service create --name my-service --host foo:127.0.0.1 --host foo:127.0.0.2 --host foo:127.0.0.3 nginx:alpine
$ docker service update --host-rm foo:127.0.0.1 --host-add foo:127.0.0.4 my-service
$ docker service inspect --format '{{.Spec.TaskTemplate.ContainerSpec.Hosts}}' my-service
[]
After this patch is applied:
$ docker service create --name my-service --host foo:127.0.0.1 --host foo:127.0.0.2 --host foo:127.0.0.3 nginx:alpine
$ docker service update --host-rm foo:127.0.0.1 --host-add foo:127.0.0.5 my-service
$ docker service inspect --format '{{.Spec.TaskTemplate.ContainerSpec.Hosts}}' my-service
[127.0.0.2 foo 127.0.0.3 foo 127.0.0.4 foo]
Signed-off-by: Sebastiaan van Stijn <github@gone.nl>
The "update" and "rollback" configurations were cross-wired, as a result, setting
`--rollback-*` options would override the service's update-options.
Creating a service with both update, and rollback configuration:
docker service create \
--name=test \
--update-failure-action=pause \
--update-max-failure-ratio=0.6 \
--update-monitor=3s \
--update-order=stop-first \
--update-parallelism=3 \
--rollback-failure-action=continue \
--rollback-max-failure-ratio=0.5 \
--rollback-monitor=4s \
--rollback-order=start-first \
--rollback-parallelism=2 \
--tty \
busybox
Before this change:
docker service inspect --format '{{json .Spec.UpdateConfig}}' test \
&& docker service inspect --format '{{json .Spec.RollbackConfig}}' test
Produces:
{"Parallelism":3,"FailureAction":"pause","Monitor":3000000000,"MaxFailureRatio":0.6,"Order":"stop-first"}
{"Parallelism":3,"FailureAction":"pause","Monitor":3000000000,"MaxFailureRatio":0.6,"Order":"stop-first"}
After this change:
{"Parallelism":3,"FailureAction":"pause","Monitor":3000000000,"MaxFailureRatio":0.6,"Order":"stop-first"}
{"Parallelism":2,"FailureAction":"continue","Monitor":4000000000,"MaxFailureRatio":0.5,"Order":"start-first"}
Signed-off-by: Sebastiaan van Stijn <github@gone.nl>
This ensures Windows paths are handled correctly as explained in the path package documentation.
Signed-off-by: Mathieu Champlon <mathieu.champlon@docker.com>
It will be helpful to expose the pull implementation which supports
pulling private images for other CLI commands that rely on helper images.
Signed-off-by: Daniel Hiltgen <daniel.hiltgen@docker.com>
- Some of them don't make sense anymore
- Some are deprecated and removed from the engine since a few versions
already.
Signed-off-by: Vincent Demeester <vincent@sbr.pm>
When updating a service with the `--force` option, the `ForceUpdate`
property of the taskspec is incremented.
Stack deploy did not take this into account, and reset this
field to its default value (0), causing the service to be
re-deployed.
This patch copies the existing value before updating the service.
Signed-off-by: Sebastiaan van Stijn <github@gone.nl>
- `replaceDockerfileForContentTrust` is only used when content trust is
enabled, so remove the boolean.
- rename `isContentTrustEnabled` to `contentTrustEnabled`
Signed-off-by: Vincent Demeester <vincent@sbr.pm>
Some of them are skipped for now (because the feature is not supported
or needs more work), some of them are fixed.
Signed-off-by: Vincent Demeester <vincent@sbr.pm>
Fix tests that failed when using cmp.Compare()
internal/test/testutil/assert
InDelta
Fix DeepEqual with kube metav1.Time
Convert some ErrorContains to assert
Signed-off-by: Daniel Nephin <dnephin@docker.com>
Commit 2b17f4c8a8 fixed the way empty labels
are taken into account (i.e. not interpolated from environment variable),
but it created a regression.
`ValidateLabel` functions doesn't allow empty label value, but it has
always been possible to pass an empty label via the cli (`docker run --label foo`).
This fixes that by not validating the label flag.
Signed-off-by: Vincent Demeester <vincent@sbr.pm>
Includes:
- [client] Remove duplicate NewClient functions
- Add API support for templated secrets and configs
- Adjust minimum API version for templated configs/secrets
Signed-off-by: Sebastiaan van Stijn <github@gone.nl>
- Add `Version` to `types.Config`
- Add a new `Services` types (that is just `[]ServiceConfig`) and add
`MarshalYAML` method on it.
- Clean other top-level custom marshaling as `Services` is the only one
required.
Signed-off-by: Vincent Demeester <vincent@sbr.pm>
Historically, the Dockerfile had to be insde the build-context, because it was
sent as part of the build-context.
3f6dc81e10
added support for passing the Dockerfile through stdin, in which case the
contents of the Dockerfile is injected into the build-context.
This patch uses the same mechanism for situations where the location of the
Dockerfile is passed, and its path is outside of the build-context.
Before this change:
$ mkdir -p myproject/context myproject/dockerfiles && cd myproject
$ echo "hello" > context/hello
$ echo -e "FROM busybox\nCOPY /hello /\nRUN cat /hello" > dockerfiles/Dockerfile
$ docker build --no-cache -f $PWD/dockerfiles/Dockerfile $PWD/context
unable to prepare context: the Dockerfile (/Users/sebastiaan/projects/test/dockerfile-outside/myproject/dockerfiles/Dockerfile) must be within the build context
After this change:
$ mkdir -p myproject/context myproject/dockerfiles && cd myproject
$ echo "hello" > context/hello
$ echo -e "FROM busybox\nCOPY /hello /\nRUN cat /hello" > dockerfiles/Dockerfile
$ docker build --no-cache -f $PWD/dockerfiles/Dockerfile $PWD/context
Sending build context to Docker daemon 2.607kB
Step 1/3 : FROM busybox
---> 6ad733544a63
Step 2/3 : COPY /hello /
---> 9a5ae1c7be9e
Step 3/3 : RUN cat /hello
---> Running in 20dfef2d180f
hello
Removing intermediate container 20dfef2d180f
---> ce1748f91bb2
Successfully built ce1748f91bb2
Signed-off-by: Sebastiaan van Stijn <github@gone.nl>
This adds the Engine version to `docker node ls`, and `.EngineVersion` as a
template option.
With this patch applied:
docker node ls
ID HOSTNAME STATUS AVAILABILITY MANAGER STATUS ENGINE VERSION
wp9231itoqsh4rqceojqo01vp * linuxkit-025000000001 Ready Active Leader 18.01.0-ce
Signed-off-by: Sebastiaan van Stijn <github@gone.nl>
backend.
Currently, the timeout is set to 0, which means no timeout. Set it to a
sane default timeout of 30 seconds.
Signed-off-by: Anusha Ragunathan <anusha.ragunathan@docker.com>
To ensure we are loading the composefile the same wether we are pointing
to swarm or kubernetes, we need to share the loading code between both.
Signed-off-by: Vincent Demeester <vincent@sbr.pm>
`--label-file` has the exact same behavior as `--env-file`, meaning any
placeholder (i.e. a simple key, no `=` sign, no value), it will get the
value from the environment variable.
For `--label-file` it should just add an empty label.
Signed-off-by: Vincent Demeester <vincent@sbr.pm>
Prefer "strict" values for orchestrator, as it's
easier to add aliases (if we think it's needed) than
to remove them later.
Signed-off-by: Sebastiaan van Stijn <github@gone.nl>
When adding a network using `docker service update --network-add`,
the new network was added by _name_.
Existing entries in a service spec are listed by network ID, which
resulted in the CLI not detecting duplicate entries for the same
network.
This patch changes the behavior to always use the network-ID,
so that duplicate entries are correctly caught.
Before this change;
$ docker network create -d overlay foo
$ docker service create --name=test --network=foo nginx:alpine
$ docker service update --network-add foo test
$ docker service inspect --format '{{ json .Spec.TaskTemplate.Networks}}' test
[
{
"Target": "9ot0ieagg5xv1gxd85m7y33eq"
},
{
"Target": "9ot0ieagg5xv1gxd85m7y33eq"
}
]
After this change:
$ docker network create -d overlay foo
$ docker service create --name=test --network=foo nginx:alpine
$ docker service update --network-add foo test
service is already attached to network foo
Signed-off-by: Sebastiaan van Stijn <github@gone.nl>
Nao YONASHIRO
Akihiro Suda
Lifubang
Kir Kolyshkin
Tõnis Tiigi
Sebastiaan van Stijn
Tibor Vass
Tibor Vass
Simon Ferquel
Sunny
Vincent Demeester
Sebastiaan van Stijn
Stephane Jeandeaux
Jean-Christophe Sirot
Ian Campbell
Silvin Lubecki
Elliot Luo
Ryan Zhang
Olli Janatuinen
Philipp Schmied
John Howard
Silvin Lubecki
Daniel Hiltgen
Jérémie Drouet
Aleksa Sarai
selansen
Li Yi
Andrew Hsu
Michael Crosby
Arash Deshmeh
Joffrey F
Xiaodong Zhang
Lihua Tang
Harald Albers
Anda Xu
mmacy
Brian Goff
Vincent Demeester
Justyn Temme
Sami Tabet
Ying Li
Derek McGowan
Eli Uriegas
Nassim 'Nass' Eddequiouaq
Zachary Romero
Mathieu Champlon
Victor Vieux
Essam A. Hassan
Bogdan Anton
Daniel Nephin
shhsu@microsoft.com
Daniel Nephin
Wataru Ishida
Anusha Ragunathan
Sungwon Han
Akim Demaille