The tag service does a `HEAD` request to get the manifest digest, where
we can then do a `GET` against the digest.
The `GET` by tag is not cacheable, but the `GET` against the digest is.
This allows proxies to work way better.
Signed-off-by: Brian Goff <cpuguy83@gmail.com>
For CI, a temporary `DOCKER_ALLOW_SCHEMA1_PUSH_DONOTUSE` environment
variable was added while we work out a solution for testing schema 1
pulls (which currently require pushing them to a local registry first
for testing).
Signed-off-by: Sebastiaan van Stijn <github@gone.nl>
Currently we show this deprecation notice for any error returned by a
registry.
Registries can return an error for any number of reasons.
Instead let's show the deprecation notice only if the fallback was
successful.
Signed-off-by: Brian Goff <cpuguy83@gmail.com>
```
distribution/push_v2_test.go:552:29: SA4021: x = append(y) is equivalent to x = y (staticcheck)
return nil, errcode.Errors(append([]error{errcode.ErrorCodeUnauthorized.WithMessage("unauthorized")}))
^
```
Signed-off-by: Sebastiaan van Stijn <github@gone.nl>
Previously we were re-using schema2.DeserializedManifest to handle oci
manifests. The issue lies in the fact that distribution started
validating the media type string during json deserialization. This
change broke our usage of that type.
Instead distribution now provides direct support for oci schemas, so use
that instead of our custom handlers.
Signed-off-by: Brian Goff <cpuguy83@gmail.com>
This commit adds the image variant to the image.(Image) type and
updates related functionality. Images built from another will
inherit the OS, architecture and variant.
Note that if a base image does not specify an architecture, the
local machine's architecture is used for inherited images. On the
other hand, the variant is set equal to the parent image's variant,
even when the parent image's variant is unset.
The legacy builder is also updated to allow the user to specify
a '--platform' argument on the command line when creating an image
FROM scratch. A complete platform specification, including variant,
is supported. The built image will include the variant, as will any
derived images.
Signed-off-by: Chris Price <chris.price@docker.com>
1.Change base.Dial to base.DailContext.
2.Remove proxyDialer that was previously used to configure a
net.Dialer to route proxy.Dialer which will route the connections
through the proxy using the connections through a SOCKS proxy.
SOCKS proxies are now supported by configuring only http.Transport.Proxy,
and no longer require changing http.Transport.Dial.
Signed-off-by: HuanHuan Ye <logindaveye@gmail.com>
Moby works perfectly when you are in a situation when one has a good and stable
internet connection. Operating in area's where internet connectivity is likely
to be lost in undetermined intervals, like a satellite connection or 4G/LTE in
rural area's, can become a problem when pulling a new image. When connection is
lost while image layers are being pulled, Moby will try to reconnect up to 5 times.
If this fails, the incompletely downloaded layers are lost will need to be completely
downloaded again during the next pull request. This means that we are using more
data than we might have to.
Pulling a layer multiple times from the start can become costly over a satellite
or 4G/LTE connection. As these techniques (especially 4G) quite common in IoT and
Moby is used to run Azure IoT Edge devices, I would like to add a settable maximum
download attempts. The maximum download attempts is currently set at 5
(distribution/xfer/download.go). I would like to change this constant to a variable
that the user can set. The default will still be 5, so nothing will change from
the current version unless specified when starting the daemon with the added flag
or in the config file.
I added a default value of 5 for DefaultMaxDownloadAttempts and a settable
max-download-attempts in the daemon config file. It is also added to the config
of dockerd so it can be set with a flag when starting the daemon. This value gets
stored in the imageService of the daemon when it is initiated and can be passed
to the NewLayerDownloadManager as a parameter. It will be stored in the
LayerDownloadManager when initiated. This enables us to set the max amount of
retries in makeDownoadFunc equal to the max download attempts.
I also added some tests that are based on maxConcurrentDownloads/maxConcurrentUploads.
You can pull this version and test in a development container. Either create a config
`file /etc/docker/daemon.json` with `{"max-download-attempts"=3}``, or use
`dockerd --max-download-attempts=3 -D &` to start up the dockerd. Start downloading
a container and disconnect from the internet whilst downloading. The result would
be that it stops pulling after three attempts.
Signed-off-by: Lukas Heeren <lukas-heeren@hotmail.com>
Signed-off-by: Sebastiaan van Stijn <github@gone.nl>
staticcheck go linter warns:
> distribution/xfer/transfer_test.go:37:2: SA2002: the goroutine calls T.Fatalf, which must be called in the same goroutine as the test (staticcheck)
What it doesn't say is why. The reason is, t.Fatalf() calls t.FailNow(),
which is expected to stop test execution right now. It does so by
calling runtime.Goexit(), which, unless called from a main goroutine,
does not stop test execution.
Anyway, long story short, if we don't care much about stopping the test
case immediately, we can just replace t.Fatalf() with t.Errorf() which
still marks the test case as failed, but won't stop it immediately.
This patch was tested to check that the test fails if any of the
goroutines call t.Errorf():
1. Failure in DoFunc ("transfer function not started ...") was tested by
decreading the NewTransferManager() argument:
- tm := NewTransferManager(5)
+ tm := NewTransferManager(2)
2. Failure "got unexpected progress value" was tested by injecting a random:
- if present && p.Current <= val {
+ if present && p.Current <= val || rand.Intn(100) > 80 {
3. Failure in DoFunc ("too many jobs running") was tested by increasing
the NewTransferManager() argument:
- tm := NewTransferManager(concurrencyLimit)
+ tm := NewTransferManager(concurrencyLimit + 1)
While at it:
* fix/amend some error messages
* use _ for unused arguments of DoFunc
Signed-off-by: Kir Kolyshkin <kolyshkin@gmail.com>
Fixes linter warnings like this one:
> distribution/pull_v2.go:229:39: SA1019: os.SEEK_SET is deprecated: Use io.SeekStart, io.SeekCurrent, and io.SeekEnd. (staticcheck)
Signed-off-by: Kir Kolyshkin <kolyshkin@gmail.com>
opts/env_test: suppress a linter warning
this one:
> opts/env_test.go:95:4: U1000: field `err` is unused (unused)
> err error
> ^
Signed-off-by: Kir Kolyshkin <kolyshkin@gmail.com>
Signed-off-by: Sebastiaan van Stijn <github@gone.nl>
Format the source according to latest goimports.
Signed-off-by: Kir Kolyshkin <kolyshkin@gmail.com>
Signed-off-by: Sebastiaan van Stijn <github@gone.nl>
The warning on pull was incorrectly asking to contact registry admins.
It is kept on push however.
Pulling manifest lists with v2 schema1 manifests will not be supported thus
there is a warning for those, but wording changed to suggest repository author
to upgrade.
Finally, a milder warning on regular pull is kept ONLY for DockerHub users
in order to incite moving away from schema1.
Signed-off-by: Tibor Vass <tibor@docker.com>
This reverts commit 98fc09128b in order to
keep registry v2 schema1 handling and libtrust-key-based engine ID.
Because registry v2 schema1 was not officially deprecated and
registries are still relying on it, this patch puts its logic back.
However, registry v1 relics are not added back since v1 logic has been
removed a while ago.
This also fixes an engine upgrade issue in a swarm cluster. It was relying
on the Engine ID to be the same upon upgrade, but the mentioned commit
modified the logic to use UUID and from a different file.
Since the libtrust key is always needed to support v2 schema1 pushes,
that the old engine ID is based on the libtrust key, and that the engine ID
needs to be conserved across upgrades, adding a UUID-based engine ID logic
seems to add more complexity than it solves the problems.
Hence reverting the engine ID changes as well.
Signed-off-by: Tibor Vass <tibor@docker.com>
It has been declared deprecated by the author, and has a knack for
false-positives (as well as giving bad advice when it comes to APIs --
which is quite clear when looking at "nolint: interfacer" comments).
Signed-off-by: Aleksa Sarai <asarai@suse.de>
As people are using the UUID in `docker info` that was based on the v1 manifest signing key, replace
with a UUID instead.
Remove deprecated `--disable-legacy-registry` option that was scheduled to be removed in 18.03.
Signed-off-by: Justin Cormack <justin.cormack@docker.com>
When a manifest list contains both Linux and Windows images, always
prefer Windows when the platform OS is unspecified. Also, filter out any
Windows images with a higher build than the host, since they cannot run.
Signed-off-by: John Stephens <johnstep@docker.com>
OCI types are backwards compatible with Docker manifest
types, however the media types must be registered.
Signed-off-by: Derek McGowan <derek@mcgstyle.net>
Signed-off-by: John Howard <jhoward@microsoft.com>
Addresses https://github.com/moby/moby/pull/35089#issuecomment-367802698.
This change enables the daemon to automatically select an image under LCOW
that can be used if the API doesn't specify an explicit platform.
For example:
FROM supertest2014/nyan
ADD Dockerfile /
And docker build . will download the linux image (not a multi-manifest image)
And similarly docker pull ubuntu will match linux/amd64
vendored distribution is quite old, and current distribution contains an
API break, which means it's not possible to vendor a bugfixed
distribution and a docker/docker at the same time.
Signed-off-by: Mike Lundy <mike@fluffypenguin.org>
Since Go 1.7, context is a standard package. Since Go 1.9, everything
that is provided by "x/net/context" is a couple of type aliases to
types in "context".
Many vendored packages still use x/net/context, so vendor entry remains
for now.
Signed-off-by: Kir Kolyshkin <kolyshkin@gmail.com>
error
local digest cache will be removed when error occured on push image
but it should not be removed if it is an auth error while on auth was
provided
https://github.com/moby/moby/issues/36309
Signed-off-by: 慕陶 <jihui.xjh@alibaba-inc.com>
Signed-off-by: John Howard <jhoward@microsoft.com>
The re-coalesces the daemon stores which were split as part of the
original LCOW implementation.
This is part of the work discussed in https://github.com/moby/moby/issues/34617,
in particular see the document linked to in that issue.
Instead of having to create a bunch of custom error types that are doing
nothing but wrapping another error in sub-packages, use a common helper
to create errors of the requested type.
e.g. instead of re-implementing this over and over:
```go
type notFoundError struct {
cause error
}
func(e notFoundError) Error() string {
return e.cause.Error()
}
func(e notFoundError) NotFound() {}
func(e notFoundError) Cause() error {
return e.cause
}
```
Packages can instead just do:
```
errdefs.NotFound(err)
```
Signed-off-by: Brian Goff <cpuguy83@gmail.com>
Files that are suffixed with `_linux.go` or `_windows.go` are
already only built on Linux / Windows, so these build-tags
were redundant.
Signed-off-by: Sebastiaan van Stijn <github@gone.nl>
Signed-off-by: John Howard <jhoward@microsoft.com>
This PR has the API changes described in https://github.com/moby/moby/issues/34617.
Specifically, it adds an HTTP header "X-Requested-Platform" which is a JSON-encoded
OCI Image-spec `Platform` structure.
In addition, it renames (almost all) uses of a string variable platform (and associated)
methods/functions to os. This makes it much clearer to disambiguate with the swarm
"platform" which is really os/arch. This is a stepping stone to getting the daemon towards
fully multi-platform/arch-aware, and makes it clear when "operating system" is being
referred to rather than "platform" which is misleadingly used - sometimes in the swarm
meaning, but more often as just the operating system.
Update logic to choose manifest from manifest list to check
for os version on Windows. Separate the logic for windows
and unix to keep unix logic the same.
Signed-off-by: Derek McGowan <derek@mcgstyle.net>
Changed push message from "The push refers to a repository" to "The push refers to the repository" to make it clearer that this is just an information about the local source and not a distinction between pushing single images or whole repositories.
Signed-off-by: Pascal <pascalgn@users.noreply.github.com>
Use strongly typed errors to set HTTP status codes.
Error interfaces are defined in the api/errors package and errors
returned from controllers are checked against these interfaces.
Errors can be wraeped in a pkg/errors.Causer, as long as somewhere in the
line of causes one of the interfaces is implemented. The special error
interfaces take precedence over Causer, meaning if both Causer and one
of the new error interfaces are implemented, the Causer is not
traversed.
Signed-off-by: Brian Goff <cpuguy83@gmail.com>
Changes most references of syscall to golang.org/x/sys/
Ones aren't changes include, Errno, Signal and SysProcAttr
as they haven't been implemented in /x/sys/.
Signed-off-by: Christopher Jones <tophj@linux.vnet.ibm.com>
[s390x] switch utsname from unsigned to signed
per 33267e036f
char in s390x in the /x/sys/unix package is now signed, so
change the buildtags
Signed-off-by: Christopher Jones <tophj@linux.vnet.ibm.com>
The --allow-nondistributable-artifacts daemon option specifies
registries to which foreign layers should be pushed. (By default,
foreign layers are not pushed to registries.)
Additionally, to make this option effective, foreign layers are now
pulled from the registry if possible, falling back to the URLs in the
image manifest otherwise.
This option is useful when pushing images containing foreign layers to a
registry on an air-gapped network so hosts on that network can pull the
images without connecting to another server.
Signed-off-by: Noah Treuhaft <noah.treuhaft@docker.com>
If a reference passed to the pull code contains both a tag and a digest,
currently the tag is used instead of the digest in the request to the
registry. This is the wrong behavior. Change it to favor the digest.
Signed-off-by: Aaron Lehmann <aaron.lehmann@docker.com>
The diff id resolution currently relies on a stored mapping for
archive digest to diff id. This mapping could be derived from
the image configuration if the image configuration is available.
On linux the image config is pulled in parallel and may not be
available. On windows, however, it is always pulled first and can
be used to supplement the stored mapping for images which may not
have this mapping from being side loaded. This becomes useful when
combined with side loaded foreign layers.
Signed-off-by: Derek McGowan <derek@mcgstyle.net>
Without this fix the error the client might see is:
target is unknown
which wasn't helpful to me when I saw this today. With this fix I
now see:
MediaType is unknown: 'text/html'
which helped me track down the issue to the registry I was talking to.
Signed-off-by: Doug Davis <dug@us.ibm.com>
Per request for more debug info on how the engine deals with
multi-platform "manifest list" images, this adds information about the
manifest list entries and whether it found an os/arch match, and the
digest of the match.
Docker-DCO-1.1-Signed-off-by: Phil Estes <estesp@linux.vnet.ibm.com>
Fallback errors are not an error, but an
informational message.
This changes those errors to be logged
as "Info" instead of "Error".
After this patch, debug logs look like this;
DEBU[0050] Calling GET /_ping
DEBU[0050] Calling POST /v1.27/images/create?fromImage=localhost%3A5000%2Ffoo&tag=latest
DEBU[0050] Trying to pull localhost:5000/foo from https://localhost:5000 v2
WARN[0050] Error getting v2 registry: Get https://localhost:5000/v2/: http: server gave HTTP response to HTTPS client
INFO[0050] Attempting next endpoint for pull after error: Get https://localhost:5000/v2/: http: server gave HTTP response to HTTPS client
DEBU[0050] Trying to pull localhost:5000/foo from http://localhost:5000 v2
INFO[0050] Attempting next endpoint for pull after error: manifest unknown: manifest unknown
DEBU[0050] Trying to pull localhost:5000/foo from https://localhost:5000 v1
DEBU[0050] attempting v1 ping for registry endpoint https://localhost:5000/v1/
DEBU[0050] Fallback from error: Get https://localhost:5000/v1/_ping: http: server gave HTTP response to HTTPS client
INFO[0050] Attempting next endpoint for pull after error: Get https://localhost:5000/v1/_ping: http: server gave HTTP response to HTTPS client
DEBU[0050] Trying to pull localhost:5000/foo from http://localhost:5000 v1
DEBU[0050] [registry] Calling GET http://localhost:5000/v1/repositories/foo/images
ERRO[0050] Not continuing with pull after error: Error: image foo:latest not found
Signed-off-by: Sebastiaan van Stijn <github@gone.nl>
Remove forked reference package. Use normalized named values
everywhere and familiar functions to convert back to familiar
strings for UX and storage compatibility.
Enforce that the source repository in the distribution metadata
is always a normalized string, ignore invalid values which are not.
Update distribution tests to use normalized values.
Signed-off-by: Derek McGowan <derek@mcgstyle.net> (github: dmcgowan)
Use resolving to repo info as the split point between the
legitimate reference package and forked reference package.
Signed-off-by: Derek McGowan <derek@mcgstyle.net> (github: dmcgowan)
As noted by #30083, the new strict checking of mediatypes misses some
cases where earlier bugs caused nonstandard mediatypes to be stored in
manifests. Two of the known cases are text/html and application/json,
which were returned by certain registries and stored by earlier versions
of Docker. Add special cases for text/html and application/json.
Signed-off-by: Aaron Lehmann <aaron.lehmann@docker.com>
This patch fixed below 4 types of code line
1. Remove unnecessary variable assignment
2. Use variables declaration instead of explicit initial zero value
3. Change variable name to underbar when variable not used
4. Add erro check and return for ignored error
Signed-off-by: Daehyeok Mun <daehyeok@gmail.com>
When storeLayer.Parent returns the parent layer, it needs to use the same logic as Get where it wraps in a describablyStoreLayer if the layer is describable. Otherwise, on Windows, this can result in pushing the foreign layers, which is not supposed to be allowed.
This fixes https://github.com/docker/docker/issues/30080.
Signed-off-by: Stefan J. Wernli <swernli@microsoft.com>
The `digest` data type, used throughout docker for image verification
and identity, has been broken out into `opencontainers/go-digest`. This
PR updates the dependencies and moves uses over to the new type.
Signed-off-by: Stephen J Day <stephen.day@docker.com>
- Make it possible to define a shorter waiting time of httputils
- Make a small hack to reduce the waiting time on distribution/xfer
Signed-off-by: Vincent Demeester <vincent@sbr.pm>
Signed-off-by: Ke Li <kel@splunk.com>
Add missing changes
Signed-off-by: Ke Li <kel@splunk.com>
User errors.New to create error
Signed-off-by: Ke Li <kel@splunk.com>
Move plugins to shared distribution stack with images.
Create immutable plugin config that matches schema2 requirements.
Ensure data being pushed is same as pulled/created.
Store distribution artifacts in a blobstore.
Run init layer setup for every plugin start.
Fix breakouts from unsafe file accesses.
Add support for `docker plugin install --alias`
Uses normalized references for default names to avoid collisions when using default hosts/tags.
Some refactoring of the plugin manager to support the change, like removing the singleton manager and adding manager config struct.
Signed-off-by: Tonis Tiigi <tonistiigi@gmail.com>
Signed-off-by: Derek McGowan <derek@mcgstyle.net>
Signed-off-by: liwenqi <vikilwq@zju.edu.cn>
update some files in the folder of distribution/xfer
Signed-off-by: liwenqi <vikilwq@zju.edu.cn>
correct again
Signed-off-by: liwenqi <vikilwq@zju.edu.cn>
Move configurations into a single file.
Abstract download manager in pull config.
Add supports for schema2 only and schema2 type checking.
Add interface for providing push layers.
Abstract image store to generically handle configurations.
Signed-off-by: Derek McGowan <derek@mcgstyle.net>
When building a manifest during a push operation, all layers must have
an associated descriptor. If a layer is missing a descriptor, that leads
to a panic.
A break inside a switch in layerAlreadyExists meant to break from the
loop surrounding the switch, but instead breaks from the switch. This
causes the loop to continue, and can overwrite the descriptor with an
empty one, leading to the panic.
Also, fix layerAlreadyExists not to abort the push when a speculative
stat on a candidate layer digest fails with an error. This could happen
in situations like a potential cross-repository mount where the user
does not have permission to access the source repository.
Signed-off-by: Aaron Lehmann <aaron.lehmann@docker.com>
Modified newV1DependencyImage and generateDependencyImages not to return
errors instead of always return nil.
Signed-off-by: Daehyeok Mun <daehyeok@gmail.com>
Move some of the optional parameters of CreateRWLayer() in a struct
called CreateRWLayerOpts. This will make it easy to add more options
arguments without having to change signature of CreateRWLayer().
Signed-off-by: Vivek Goyal <vgoyal@redhat.com>
Translate pull errors to provide a more consistent and user friendly
error message.
Signed-off-by: Derek McGowan <derek@mcgstyle.net> (github: dmcgowan)
Remove the following comment in pullV2Tag:
// NOTE: not using TagService.Get, since it uses HEAD requests
// against the manifests endpoint, which are not supported by
// all registry versions.
This is actually not an issue, because TagService.Get does a fallback to
GET if HEAD fails. It has done this ever since TagService was added to
the distribution API, so this comment was probably based on an early
version of TagService before it was merged, or was always a
misunderstanding.
However, we continue to use ManifestService.Get instead because it
saves a round trip. The manifest can be retrieved directly instead of
resolving the digest first.
Signed-off-by: Aaron Lehmann <aaron.lehmann@docker.com>
Allow built images to be squash to scratch.
Squashing does not destroy any images or layers, and preserves the
build cache.
Introduce a new CLI argument --squash to docker build
Introduce a new param to the build API endpoint `squash`
Once the build is complete, docker creates a new image loading the diffs
from each layer into a single new layer and references all the parent's
layers.
Signed-off-by: Brian Goff <cpuguy83@gmail.com>
Classify blobs into three categories based on size.
Use a very limited number of mount attempts and no existence check for
small blobs. Use more attempts for bigger blobs.
Also remember blob associations during layer existence check.
Blob digests are now checked in the target repository from newest to
latest. If the blob exists and the metadata entry does not, it will be
created. If the blob is not found, the metadata entry will be removed.
Signed-off-by: Michal Minář <miminar@redhat.com>
Don't fallback back immediately to blob upload if the cross-repo mount
fails and layer upload is initiated by registry.
Instead cancel the upload and re-try cross-repo mount from different
source repository before doing full re-upload.
Signed-off-by: Michal Minář <miminar@redhat.com>
to avoid unnecessary blob re-uploads.
Cross repository mount from particular repo will most probably fail if
the user pushing to the registry is not the same as the one who pulled
or pushed to the source repo.
This PR attempts first to cross-repo mount from the source repositories
associated with the pusher's auth config. Then it falls back to other
repositories sorted from the most similar to the target repo to the
least.
It also prevents metadata deletion in cases where cross-repo mount fails
and the auth config hashes differ.
Signed-off-by: Michal Minář <miminar@redhat.com>
