Context should use type values instead of strings.
Updated direct calls to WithValue, but still other uses of string keys.
Update Acl to ACL in s3 driver.
Signed-off-by: Derek McGowan <derek@mcgstyle.net> (github: dmcgowan)
The Redis tests were failing with a "connection pool exhausted" error
from Redigo. Closing the connection used for FLUSHDB fixes the problem.
Signed-off-by: Noah Treuhaft <noah.treuhaft@docker.com>
The Hub registry generates a large volume of notifications, many of
which are uninteresting based on target media type. Discarding them
within the notification endpoint consumes considerable resources that
could be saved by discarding them within the registry. To that end,
this change adds registry configuration options to restrict the
notifications sent to an endpoint based on target media type.
Signed-off-by: Noah Treuhaft <noah.treuhaft@docker.com>
Access logging is great. Access logging you can turn off is even
better. This change adds a configuration option for that.
Signed-off-by: Noah Treuhaft <noah.treuhaft@docker.com>
The token auth package logs JWT validation and verification failures at
the `error` level. But from the server's perspective, these aren't
errors. They're the expected response to bad input. Logging them at
the `info` level better reflects that distinction.
Signed-off-by: Noah Treuhaft <noah.treuhaft@docker.com>
If a user specifies `mydomain.com:443` in the `Host` configuration, the
PATCH request for the layer upload will fail because the challenge does not
appear to be in the map. To fix this, we normalize the map keys to always
use the Host:Port combination.
Closes https://github.com/docker/docker/issues/18469
Signed-off-by: Stan Hu <stanhu@gmail.com>
Running with the race detector may cause some parts
of the code to run slower causing a race in the scheduler
ordering.
Signed-off-by: Derek McGowan <derek@mcgstyle.net> (github: dmcgowan)
* tag service: properly handle error responses on HEAD requests by
re-issuing requests as GET for proper error details.
Fixes#1911.
Signed-off-by: dmitri <deemok@gmail.com>
* Simplify handling of failing HEAD requests in TagService and
make a GET request for cases:
- if the server does not handle HEAD
- if the response was an error to get error details
Signed-off-by: dmitri <deemok@gmail.com>
* Add a missing http.Response.Body.Close call for the GET request.
Signed-off-by: dmitri <deemok@gmail.com>
This change to the S3 Move method uses S3's multipart upload API to copy
objects whose size exceeds a threshold. Parts are copied concurrently.
The level of concurrency, part size, and threshold are all configurable
with reasonable defaults.
Using the multipart upload API has two benefits.
* The S3 Move method can now handle objects over 5 GB, fixing #886.
* Moving most objects, and espectially large ones, is faster. For
example, moving a 1 GB object averaged 30 seconds but now averages 10.
Signed-off-by: Noah Treuhaft <noah.treuhaft@docker.com>
This is already supported by ncw/swift, so we just need to pass the
parameters from the storage driver.
Signed-off-by: Stefan Majewsky <stefan.majewsky@sap.com>
Use the much faster math/rand.Read function where cryptographic
guarantees are not required. The unit test suite should speed up a
little bit but we've already optimized around this, so it may not
matter.
Signed-off-by: Stephen J Day <stephen.day@docker.com>
Previous component-wise path comparison is recursive and generates a
large amount of garbage. This more efficient version simply replaces the
path comparison with the zero-value to sort before everything. We do
this by replacing the byte-wise comparison that swaps a single character
inline for the separator comparison, such that separators sort first.
The resulting implementation provides component-wise path comparison
with no cost incurred for allocation or stack frame.
Direction of the comparison is also reversed to match Go style.
Signed-off-by: Stephen J Day <stephen.day@docker.com>
* Allow precomputed stats on cross-mounted blobs
Signed-off-by: Michal Minář <miminar@redhat.com>
* Extended cross-repo mount tests
Signed-off-by: Michal Minář <miminar@redhat.com>
* Add Object ACL Support to the S3 Storage Backend
Signed-off-by: Frank Chen <frankchn@gmail.com>
* Made changes per @RichardScothern's comments
Signed-off-by: Frank Chen <frankchn@gmail.com>
* Fix Typos
Signed-off-by: Frank Chen <frankchn@gmail.com>
Pass the manifestURL directly into the schema2 manifest handler instead of
accessing through the repository as it has since the reference is now an
interface.
Signed-off-by: Richard Scothern <richard.scothern@docker.com>
Until we have some experience hosting foreign layer manifests, the Hub
operators wish to limit foreign layers on Hub. To that end, this change
adds registry configuration options to restrict the URLs that may appear
in pushed manifests.
Signed-off-by: Noah Treuhaft <noah.treuhaft@docker.com>
Adds a constant leeway (60 seconds) to the nbf and exp claim check to
account for clock skew between the registry servers and the
authentication server that generated the JWT.
The leeway of 60 seconds is a bit arbitrary but based on the RFC
recommendation and hub.docker.com logs/metrics where we don't see
drifts of more than a second on our servers running ntpd.
I didn't attempt to make the leeway configurable as it would add extra
complexity to the PR and I am not sure how Distribution prefer to
handle runtime flags like that.
Also, I am simplifying the exp and nbf check for readability as the
previous `NOT (A AND B)` with cmp operators was not very friendly.
Ref:
https://tools.ietf.org/html/rfc7519#section-4.1.5
Signed-off-by: Marcus Martins <marcus@docker.com>
This fixes errors other than io.EOF from being dropped when a storage driver
lists repositories. For example, filesystem driver may point to a missing
directory and errors, which then gets subsequently dropped.
Signed-off-by: Edgar Lee <edgar.lee@docker.com>
Allows using v2 for v1 endpoints.
The primary use case being for search which does not have a v2 specification.
Added a user scope for allowing v2 search
Signed-off-by: Derek McGowan <derek@mcgstyle.net> (github: dmcgowan)
This is similar to waitForSegmentsToShowUp which is called during
Close/Commit. Intuitively, you wouldn't expect missing segments to be a
problem during read operations, since the previous Close/Commit
confirmed that all segments are there.
But due to the distributed nature of Swift, the read request could be
hitting a different storage node of the Swift cluster, where the
segments are still missing.
Load tests on my team's staging Swift cluster have shown this to occur
about once every 100-200 layer uploads when the Swift proxies are under
high load. The retry logic, borrowed from waitForSegmentsToShowUp, fixes
this temporary inconsistency.
Signed-off-by: Stefan Majewsky <stefan.majewsky@sap.com>
In Go's header parsing, the same header multiple times results in multiple entries in the `r.Header[...]` slice, but Go does no further parsing beyond that (and in https://golang.org/cl/4528086 it was determined that until/unless the stdlib itself needs it, Go will not do so).
The consequence here for parsing of `Accept:` headers is that we support the way Go outputs headers, but not all language HTTP libraries have a facility to output multiple headers instead of a single list header.
This change ensures that the following (valid) header blocks all parse to the same result for the purposes of what is being tested here:
```
Accept: a/b
Accept: b/c
Accept: d/e
```
```
Accept: a/b; q=0.5, b/c
Accept: d/e
```
```
Accept: a/b; q=0.1, b/c; q=0.2, d/e; q=0.8
```
Signed-off-by: Andrew "Tianon" Page <admwiggin@gmail.com>
The client may need the content digest to delete a manifest using the digest used by the registry.
Signed-off-by: Derek McGowan <derek@mcgstyle.net> (github: dmcgowan)
This lets us access registry config within middleware for additional
configuration of whatever it is that you're overriding.
Signed-off-by: Tony Holdstock-Brown <tony@docker.com>
go1.5 doesn't export http.StatusTooManyRequests while
go1.6 does. Fix this by hardcoding the status code for now.
Signed-off-by: Antonio Murdaca <runcom@redhat.com>
This commit refactors base.regulator into the 2.4 interfaces and adds a
filesystem configuration option `maxthreads` to configure the regulator.
By default `maxthreads` is set to 100. This means the FS driver is
limited to 100 concurrent blocking file operations. Any subsequent
operations will block in Go until previous filesystem operations
complete.
This ensures that the registry can never open thousands of simultaneous
threads from os filesystem operations.
Note that `maxthreads` can never be less than 25.
Add test case covering parsable string maxthreads
Signed-off-by: Tony Holdstock-Brown <tony@docker.com>
subsequent close.
When a blob upload is cancelled close the blobwriter before removing
upload state to ensure old hashstates don't persist.
Signed-off-by: Richard Scothern <richard.scothern@docker.com>
It's easily possible for a flood of requests to trigger thousands of
concurrent file accesses on the storage driver. Each file I/O call creates
a new OS thread that is not reaped by the Golang runtime. By limiting it
to only 100 at a time we can effectively bound the number of OS threads
in use by the storage driver.
Docker-DCO-1.1-Signed-off-by: Josh Hawn <josh.hawn@docker.com> (github: jlhawn)
Signed-off-by: Tony Holdstock-Brown <tony@docker.com>
Not just when Commit()ing the result. This fixes some errors I observed
when the layer (i.e. the DLO) is Stat()ed immediately after closing,
and reports the wrong file size because the container listing is not
yet up-to-date.
Signed-off-by: Stefan Majewsky <stefan.majewsky@sap.com>
If a schema 1 manifest is uploaded with the `disablesignaturestore` option set
to true, then no signatures will exist. Handle this case.
If a schema 1 manifest is pushed, deleted, garbage collected and pushed again, the
repository will contain signature links from the first version, but the blobs will
not exist. Disable the signature store in the garbage-collect command so
signatures are not fetched.
Signed-off-by: Richard Scothern <richard.scothern@docker.com>
In 326c3a9c49, which was only intended to
be a refactoring commit, the behavior of this block subtly changed so
that unknown types of errors would be swallowed instead of propagated.
I noticed this while investigating an error similar to #1539 aka
docker/docker#21290. It appears that during GetContent() for a
hashstate, the Swift proxy produces an error. Since this error was
silently swallowed, an empty []byte is used to restart the hash, then
producing the digest of the empty string instead of the layer's digest.
This PR will not fix the issue, but it should make the actual error more
visible by propagating it into `blobWriter#resumeDigest' and
'blobWriter#validateBlob', respectively.
Signed-off-by: Stefan Majewsky <stefan.majewsky@sap.com>
This commit adds context-specific documentation on StorageDriver,
StorageDriverFactory, and the factory’s Register func, explaining how
the internal registration mechanism should be used.
This documentation follows from the thread starting at
https://github.com/deis/builder/pull/262/files#r56720200.
cc/ @stevvooe
Signed-off-by: Aaron Schlesinger <aschlesinger@deis.com>
If this slice ends up empty after parsing the HTTP response body, it
means the body is not well-formed. We've probably encountered an error
message produced by something that uses a different JSON schema, or
an error that just happens to validate as JSON.
An empty errcode.Errors slice is not a very useful thing to return,
since its Error() output is just `<nil>`. Detect this case, and instend
return an UnexpectedHTTPResponseError.
Signed-off-by: Aaron Lehmann <aaron.lehmann@docker.com>
Updates registry storage code to use this for better resumable writes.
Implements this interface for the following drivers:
+ Inmemory
+ Filesystem
+ S3
+ Azure
Signed-off-by: Brian Bland <brian.bland@docker.com>
It is possible for a middlebox to lowercase the URL at somepoint causing a
lookup in the auth challenges table to fail. Lowercase hostname before
using as keys to challenge map.
Signed-off-by: Richard Scothern <richard.scothern@gmail.com>
Cross repository push tokens were not being cached and could not be used,
now any returned token will be used and the caching is hidden in the getToken function.
Signed-off-by: Derek McGowan <derek@mcgstyle.net> (github: dmcgowan)
Login needs to add an offline token flag to ensure a refresh token is returned by the token endpoint.
Signed-off-by: Derek McGowan <derek@mcgstyle.net> (github: dmcgowan)
The Move operation is only used to move uploaded blobs
to their final destination. There is no point in implementing
Move on "folders". Apart from simplifying the code, this also
saves an HTTP request.
Signed-off-by: Arthur Baars <arthur@semmle.com>
- Includes a change in the command to run the registry. The registry
server itself is now started up as a subcommand.
- Includes changes to the high level interfaces to support enumeration
of various registry objects.
Signed-off-by: Andrew T Nguyen <andrew.nguyen@docker.com>
Ensure Accept headers are sent with TagService.Get (which hits manifest
endpoints). Add support for remote Get and Put for the proxied blobstore.
Signed-off-by: Richard Scothern <richard.scothern@gmail.com>
Middleware code may perform additional checks on blobs written. Allow it
to return access denied errors that will result in 403 Forbidden.
Signed-off-by: Michal Minar <miminar@redhat.com>
Offset can be more than CurrentSize as long as this case is checked
by DriverSuite.testContinueStreamAppend.
Signed-off-by: Anton Tiurin <noxiouz@yandex.ru>
Fixes bug in TestStorageClass for s3aws driver where the "standard" file
was checked for reduced-redundnancy storage.
Signed-off-by: Brian Bland <brian.bland@docker.com>
challenges with the upstream until any proxied data is found not to be local.
Implement auth challenges behind an interface and add to unit tests. Also,
remove a non-sensical unit test.
Signed-off-by: Richard Scothern <richard.scothern@docker.com>
Keeps old s3 driver, renames to s3goamz, registers new s3 driver as both
"s3" and "s3aws"
Changes cloudfront middleware to use aws-sdk-go
Signed-off-by: Brian Bland <brian.bland@docker.com>
Making this an exported error value will allow users of the
registry/client/auth module to have consistent behavior between
authentication failures and cases where no credentials are provided.
Signed-off-by: Aaron Lehmann <aaron.lehmann@docker.com>
Add option for specifying trust key for signing schema1 manifests.
Since schema1 signature key identifiers are not verified anywhere and deprecated, storing signatures is no longer a requirement.
Furthermore in schema2 there is no signature, requiring the registry to already add signatures to generated schema1 manifests.
Signed-off-by: Derek McGowan <derek@mcgstyle.net> (github: dmcgowan)
When schema2 manifests are rewritten as schema1 currently the etag and docker content digest header keep the value for the schema2 manifest.
Fixes#1444
Signed-off-by: Derek McGowan <derek@mcgstyle.net> (github: dmcgowan)
Treats nil parameters the same as unprovided parameters (fixes issues
where certain parameters are printed to "<nil>").
Accepts "true" and "false" string values for boolean parameters.
Signed-off-by: Brian Bland <brian.bland@docker.com>
A changeset under consideration for Go 1.7 would automatically copy
headers on redirect. This change future-proofs our code so we won't make
duplicate copies of the headers if net/http does it automatically in the
future.
Signed-off-by: Aaron Lehmann <aaron.lehmann@docker.com>
Uses docker/goamz instead of AdRoll/goamz
Adds a registry UA string param to the storage parameters when
constructing the storage driver for the registry App.
This could be used by other storage drivers as well
Signed-off-by: Brian Bland <brian.bland@docker.com>
Prevent using strings throughout the code to reference a string key defined in the auth package.
Signed-off-by: Derek McGowan <derek@mcgstyle.net> (github: dmcgowan)
Since RawMessage json receivers take a pointer type, the Header structure should use points in order to call the json.RawMessage marshal and unmarshal functions
Signed-off-by: Derek McGowan <derek@mcgstyle.net> (github: dmcgowan)
Most places in the registry were using string types to refer to
repository names. This changes them to use reference.Named, so the type
system can enforce validation of the naming rules.
Signed-off-by: Aaron Lehmann <aaron.lehmann@docker.com>
This is needed for compatibility with some third-party registries that
send an inappropriate Content-Type header such as text/html.
Signed-off-by: Aaron Lehmann <aaron.lehmann@docker.com>
https://github.com/docker/distribution/pull/1249 changed token fetching
to parse HTTP error response bodies as serialized errcodes. However,
Docker Hub's authentication endpoint does not return error bodies in
this format. To work around this, convert its format into
ErrCodeUnauthorized or ErrCodeUnknown.
Signed-off-by: Aaron Lehmann <aaron.lehmann@docker.com>
Verify that the file(s) have been deleted after calling Delete,
and retry if this is not the case. Furthermore, report the error
if a Delete operation fails.
Signed-off-by: Arthur Baars <arthur@semmle.com>
The Payload function for schema1 currently returns a signed manifest,
but indicates the content type is that of a manifest that isn't signed.
Note that this breaks compatibility with Registry 2.3 alpha 1 and
Docker 1.10-rc1, because they use the incorrect content type.
Signed-off-by: Aaron Lehmann <aaron.lehmann@docker.com>
For compatibility with other registries that don't use this exact
variant of the Content-Type header, we need to be more flexible about
what we accept. Any form of "application/json" should be allowed. The
charset should not be included in the comparison.
See docker/docker#19400.
Signed-off-by: Aaron Lehmann <aaron.lehmann@docker.com>
Accidentally checked for err != nil instead of err == nil :/
Also now ensures that only a non-nil option is appended to the create
options slice
Signed-off-by: Brian Bland <brian.bland@docker.com>
Remove the requirement of file system access to run GCS unit tests. Deconstruct
the input parameters to take the private key and email which can be specified on
the build system via environment variables.
Signed-off-by: Richard Scothern <richard.scothern@gmail.com>
Removes the Mount operation and instead implements this behavior as part
of Create a From option is provided, which in turn returns a rich
ErrBlobMounted indicating that a blob upload session was not initiated,
but instead the blob was mounted from another repository
Signed-off-by: Brian Bland <brian.bland@docker.com>
When an auth request provides the "from" query parameter, the token
handler will add a "pull" scope for the provided repository, refreshing
the token if the overall scope has increased
Signed-off-by: Brian Bland <brian.bland@docker.com>
This makes content type sniffing cleaner. The document just needs to be
decoded into a manifest.Versioned structure. It's no longer a two-step
process.
Signed-off-by: Aaron Lehmann <aaron.lehmann@docker.com>
Create signedManifestHandler and schema2ManifestHandler. Use these to
unmarshal and put the respective types of manifests from manifestStore.
Signed-off-by: Aaron Lehmann <aaron.lehmann@docker.com>
When a manifest is deleted by digest, look up the referenced tags in the tag
store and remove all associations.
Signed-off-by: Richard Scothern <richard.scothern@gmail.com>
Add a generic Manifest interface to represent manifests in the registry and
remove references to schema specific manifests.
Add a ManifestBuilder to construct Manifest objects. Concrete manifest builders
will exist for each manifest type and implementations will contain manifest
specific data used to build a manifest.
Remove Signatures() from Repository interface.
Signatures are relevant only to schema1 manifests. Move access to the signature
store inside the schema1 manifestStore. Add some API tests to verify
signature roundtripping.
schema1
-------
Change the way data is stored in schema1.Manifest to enable Payload() to be used
to return complete Manifest JSON from the HTTP handler without knowledge of the
schema1 protocol.
tags
----
Move tag functionality to a seperate TagService and update ManifestService
to use the new interfaces. Implement a driver based tagService to be backward
compatible with the current tag service.
Add a proxyTagService to enable the registry to get a digest for remote manifests
from a tag.
manifest store
--------------
Remove revision store and move all signing functionality into the signed manifeststore.
manifest registration
---------------------
Add a mechanism to register manifest media types and to allow different manifest
types to be Unmarshalled correctly.
client
------
Add ManifestServiceOptions to client functions to allow tags to be passed into Put and
Get for building correct registry URLs. Change functional arguments to be an interface type
to allow passing data without mutating shared state.
Signed-off-by: Richard Scothern <richard.scothern@gmail.com>
Signed-off-by: Richard Scothern <richard.scothern@docker.com>
tarsum is not actually used by the registry. Remove support for it.
Convert numerous uses in unit tests to SHA256.
Update docs to remove mentions of tarsums (which were often inaccurate).
Remove tarsum dependency.
Signed-off-by: Aaron Lehmann <aaron.lehmann@docker.com>
The current implementation of digest.FromBytes returns an error. This
error can never be non-nil, but its presence in the function signature
means each call site needs error handling code for an error that is
always nil.
I verified that none of the hash.Hash implementations in the standard
library can return an error on Write. Nor can any of the hash.Hash
implementations vendored in distribution.
This commit changes digest.FromBytes not to return an error. If Write
returns an error, it will panic, but as discussed above, this should
never happen.
This commit also avoids using a bytes.Reader to feed data into the hash
function in FromBytes. This makes the hypothetical case that would panic
a bit more explicit, and should also be more performant.
Signed-off-by: Aaron Lehmann <aaron.lehmann@docker.com>
This calls Stat before Open, which should be unnecessary because Open
can handle the case of a nonexistent blob. Removing the Stat saves a
round trip.
This is similar to the removal of stat in Open in #1226.
Signed-off-by: Aaron Lehmann <aaron.lehmann@docker.com>
This fixes both the s3 driver and the oss driver to return the unmunged path
when returning errors.
Signed-off-by: Stephen J Day <stephen.day@docker.com>
Swift returns an empty object list when trying to read a non-existing object path, treat it as a
PathNotFoundError when trying to list a non existing virtual directory.
Signed-off-by: David li <wenquan.li@hpe.com>
RADOS returns a -EIO when trying to read a non-existing OMAP, treat it as a
PathNotFoundError when trying to list a non existing virtual directory.
Signed-off-by: Vincent Giersch <vincent@giersch.fr>
Issue #1186 describes a condition where a null tags response is returned when
using the s3 driver. The issue seems to be related to a missing
PathNotFoundError in s3. This change adds a test for that to get an idea of the
lack of compliance across storage drivers. If the failures are manageable,
we'll add this test condition and fix the s3 driver.
Signed-off-by: Stephen J Day <stephen.day@docker.com>
This change removes the sort() from the Repositories() function since
we're now guaranteed to have a lexigraphically sorted walk.
Signed-off-by: Patrick Devine <patrick.devine@docker.com>
Without this commit, three round-trips are required to fetch a blob with
a progress bar. The first is a call to Stat (HEAD request), to determine
the size. Then Open is called, which also calls Stat, and finally
performs a GET request.
Only the GET request is actually needed. The size of the blob can be
sniffed from Content-Length in the GET response.
This commit changes HTTPReadSeeker to automatically detect the size from
Content-Length instead of requiring it to be passed in. The Stat call is
removed from Open because it is no longer necessary.
HTTPReadSeeker now takes an additional errorHandler callback argument which
translates an unsuccessful HTTP response into an appropriate API-level
error. Using a callback for this makes it possible to avoid leaking the
repsonse body to Read's caller, which would make lifecycle management
problematic.
Fixes#1223
Signed-off-by: Aaron Lehmann <aaron.lehmann@docker.com>
The Delete method lists objects under the given path and tries to delete
all of them with a bulk delete request. If the path has no objects
underneath it, the body of this request will be empty, which causes
HTTP-level issues. Specifically, Go's HTTP client senses the empty
request buffer and doesn't include a Content-Length, which causes the
Swift server to fail the request.
This commit fixes the problem by avoiding sending empty bulk delete
requests. This is the correct thing to do anyway, since there's no
reason to request deletion of zero objects.
Signed-off-by: Aaron Lehmann <aaron.lehmann@docker.com>
Previously, this error message would stringify as a pointer address,
which isn't particularly helpful.
This change breaks out the elements of the challenge object such that
the error is appropriately represented.
Signed-off-by: Ted Reed <ted.reed@gmail.com>
contains equal length History and FSLayer arrays.
This is required to prevent malformed manifests being put to the registry and
failing external verification checks.
Signed-off-by: Richard Scothern <richard.scothern@gmail.com>
Errors thrown by storage drivers don't have the name of the driver, causing user
confusion about whether the error is coming from Docker or from a storage driver.
This change adds the storage driver name to each error message.
This required changing ErrUnsupportedDriver to a type, leading to code changes
whenever ErrUnsupportedDriver is used. The tests check whether the driver name
appears in the error message.
Signed-off-by: Amit Shukla <amit.shukla@docker.com>
This extends the specification for the Bearer token response to include
information pertaining to when an issued Bearer token will expire.
This also allows the client to accept `access_token` as an alias for `token`.
Signed-off-by: Matt Moore <mattmoor@google.com>
The proxy scheduler implemented its own timer state machine. It's
simpler and more efficient to leverage the Go runtime's timer heap by
using time.AfterFunc.
This commit adds a time.Timer to each scheduler entry, and starts and
stops those timers as necessary. Then the mainloop goroutine and its
associated logic are not needed.
Signed-off-by: Aaron Lehmann <aaron.lehmann@docker.com>
To ensure that we only unmarshal the verified payload into the contained
manifest, we first copy the entire incoming buffer into Raw and then unmarshal
only the Payload portion of the incoming bytes. If the contents is later
verified, the caller can then be sure that the contents of the Manifest fields
can be trusted.
Signed-off-by: Aaron Lehmann <aaron.lehmann@docker.com>
There seems to be a need for a type that represents a way of pointing
to an image, irrespective of the implementation.
This patch defines a Reference interface and provides 3 implementations:
- TagReference: when only a tag is provided
- DigestReference: when a digest (according to the digest package) is
provided, can include optional tag as well
Validation of references are purely syntactic.
There is also a strong type for tags, analogous to digests, as well
as a strong type for Repository from which clients can access the
hostname alone, or the repository name without the hostname, or both
together via the String() method.
For Repository, the files names.go and names_test.go were moved from
the v2 package.
Signed-off-by: Tibor Vass <tibor@docker.com>
of readers joining current downloads. Concurrent requests for the same blob
will not block, but only the first instance will be comitted locally.
Signed-off-by: Richard Scothern <richard.scothern@gmail.com>
Fixes https://github.com/docker/distribution/issues/1062
This relaxes the naming restrictions places on Docker images to permit
valid hostnames according to [RFC-2396](https://www.ietf.org/rfc/rfc2396.txt).
It deviates from the RFC in the following ways:
1) Allow underscores where we allow hyphens (hostnames don't allow
underscores, which we must for backwards compatibility).
2) Leave "top-level" name segments unrestricted (domains require an
alpha character to begin a top-level domain, e.g. "com").
3) DO NOT allow a trailing dot, as permitted by FQDNs.
Signed-off-by: Matt Moore <mattmoor@google.com>
This allows the administrator to specify an externally-reachable URL for
the registry. It takes precedence over the X-Forwarded-Proto and
X-Forwarded-Host headers, and the hostname in the request.
Signed-off-by: Aaron Lehmann <aaron.lehmann@docker.com>
Associate HTTP 401s with Authentication errors rather than Authorization
errors. Changes the meaning of the UNAUTHORIZED error to be authentication
specific.
Defines DENIED error code to be associated with authorization
errors which result in HTTP 403 responses.
Add 'No Such Repository' errors to more endpoints.
Docker-DCO-1.1-Signed-off-by: Josh Hawn <josh.hawn@docker.com> (github: jlhawn)
The "testing" package adds some flags in its init function, so utilities
that import distribution code may print a page of extra testing flags in
their help output.
This commit solves the issue by moving an import of "testing" in the
registry/storage/cache package to a new
registry/storage/cache/cachecheck package, which is only imported by
tests.
Signed-off-by: Aaron Lehmann <aaron.lehmann@docker.com>
Add "readonly" under the storage/maintenance section. When this is set
to true, uploads and deletions will return 503 Service Unavailable
errors.
Document the parameter and add some unit testing.
Signed-off-by: Aaron Lehmann <aaron.lehmann@docker.com>
In the S3 storage driver there is currently an initial access permission check by listing the bucket. If this check fails, registry will panic and exit.
However, this check is broken in two ways. First of all it strips the final slash from the root directory path, meaning that any access permissions which limit access to a single directory will fail, because S3 treats the path as strict prefix match. Secondly it fails to strip any leading slash that might be present, unlike the other access places, which means that the path used is different as a leading slash is allowed and significant in a filename in S3.
Since there is also a periodic health check which correctly checks access permissions and shows the error more cleanly, the best solution seems to be to just remove this initial access check.
Signed-off-by: Nuutti Kotivuori <nuutti.kotivuori@poplatek.fi>
By adding WithVersion to the context package, we can simplify context setup in
the application. This avoids some odd bugs where instantiation order can lead
to missing instance.id or version from log messages.
Signed-off-by: Stephen J Day <stephen.day@docker.com>
This solves a issue from #909 where instance.id was not printed in logs,
because this file was using the background context from
golang.org/x/net/context instead of
github.com/docker/distribution/context.
It's cleaner to standardize on one package, so this commit removes the
import of golang.org/x/net/context entirely. The Context interfaces
defined in both packages are the same, so other code using
golang.org/x/net/context can still pass its context to NewRegistry.
Signed-off-by: Aaron Lehmann <aaron.lehmann@docker.com>
Use this command in cmd/registry/main.go.
Move debug server to the main command, and change Serve to be a
ListenAndServe function.
Signed-off-by: Aaron Lehmann <aaron.lehmann@docker.com>
As we begin our march towards multi-arch, we must prepare for the reality of
multiple manifest schemas. This is the beginning of a set of changes to
facilitate this. We are both moving this package into its target position where
it may live peacefully next to other manfiest versions.
Signed-off-by: Stephen J Day <stephen.day@docker.com>
To bring ErrorCode into liine with Go conventions, ErrorCode.Error() now
returns the "nice" value of the error code. This ensures error message assembly
works similar to commonly used Go conventions when directly using ErrorCode as
an error.
Signed-off-by: Stephen J Day <stephen.day@docker.com>
Also, add timeout and status code parameters to the HTTP checker, and
remove the threshold parameter for the file checker.
Signed-off-by: Aaron Lehmann <aaron.lehmann@docker.com>
Update docs.
Change health_test.go tests to create their own registries and register
the checks there. The tests now call CheckStatus directly instead of
polling the HTTP handler, which returns results from the default
registry.
Signed-off-by: Aaron Lehmann <aaron.lehmann@docker.com>
Add default storagedriver health check to example configuration files
with parameters matching the previous hardcoded configuration.
Signed-off-by: Aaron Lehmann <aaron.lehmann@docker.com>
Add a section to the config file called "health". Within this section,
"filecheckers" and "httpcheckers" list checks to run. Each check
specifies a file or URI, a time interval for the check, and a threshold
specifying how many times the check must fail to reach an unhealthy
state.
Document the new options in docs/configuration.md.
Add unit testing for both types of checkers. Add an UnregisterAll
function in the health package to support the unit tests, and an
Unregister function for consistency with Register.
Fix a string conversion problem in the health package's HTTP checker.
Signed-off-by: Aaron Lehmann <aaron.lehmann@docker.com>
Clean up calling convention for NewRegistryWithDriver to use functional
arguments.
This is a first step towards the refactor described in #215. I plan to
add additional options in the process of moving configurable items from
the App structure to the registry structure.
Signed-off-by: Aaron Lehmann <aaron.lehmann@docker.com>
A change in #763 to address review comments caused problems. Originally,
instrumentedResponseWriter implemented the CloseNotifier interface, and
would panic if it was wrapping something that did not implement that
interface. This was split into a separate instrumentedResponseWriterCN
type that implements CloseNotifier, so there's a fallback if
instrumentedResponseWriter ever needs to wrap something that does not
implement this interface.
instrumentedResponseWriter's Value method would end up upcasting either
type back to instrumentedResponseWriter, which does not implement the
interface. In effect, instrumentedResponseWriterCN was never visible to
the handler.
This fixes the problem by implementing a wrapper Value method for
instrumentedResponseWriterCN.
Signed-off-by: Aaron Lehmann <aaron.lehmann@docker.com>
Change checkResponse to only expect the configured
X-Content-Type-Options header if it doesn't receive a 405 error, which
means the handler isn't registered for that method.
Signed-off-by: Aaron Lehmann <aaron.lehmann@docker.com>
The use of the pathMapper is no longer needed the way we have organized the
code base. The extra level of indirection has proved unnecessary and confusing
so we've opted to clean it up. In the future, we may require more flexibility,
but now it is simply not required.
Signed-off-by: Stephen J Day <stephen.day@docker.com>
Unfortunately, the 2.1 releease has written manfiest links into the wrong
directory. This doesn't affect new 2.1 deployments but fixing this to be 2.0
backwards compatible has broken 2.1.0 compatibility. To ensure we have
compatibility between 2.0, 2.1.0 and future releases, we now check one of
several locations to identify a manifest link.
Signed-off-by: Stephen J Day <stephen.day@docker.com>
Unfortunately, the refactor used the incorrect path for manifest links within a
repository. While this didn't stop the registry from working, it did break
compatibility with 2.0 deployments for manifest fetches.
Tests were added to ensure these are locked down to the appropriate paths.
Signed-off-by: Stephen J Day <stephen.day@docker.com>
The example configuration files add X-Content-Type-Options: nosniff.
Add coverage in existing registry/handlers unit tests.
Signed-off-by: Aaron Lehmann <aaron.lehmann@docker.com>
distribution errors. Fill in missing checks for mutations on a registry pull-through
cache. Add unit tests and update documentation.
Also, give v2.ErrorCodeUnsupported an HTTP status code, previously it was
defaulting to 500, now its 405 Method Not Allowed.
Signed-off-by: Richard Scothern <richard.scothern@gmail.com>
Several error codes are generally useful but tied to the v2 specification
definitions. This change moves these error code definitions into the common
package for use by the health package, which is not tied to the v2 API.
Signed-off-by: Stephen J Day <stephen.day@docker.com>
To ensure the ensure the web application is properly operating, we've added a
periodic health check for the storage driver. If the health check fails three
times in a row, the registry will serve 503 response status for any request
until the condition is resolved. The condition is reported in the response body
and via the /debug/health endpoint.
To ensure that all drivers will properly operate with this health check, a
function has been added to the driver testsuite.
Signed-off-by: Stephen J Day <stephen.day@docker.com>
When using the RADOS driver, the hierarchy of the files is stored
in OMAPs, but the root OMAP was not created and a call to List("/")
was returning an error instead of returned the first level files
stored. This patches creates an OMAP for "/" and excludes the listed
directory from the list of files returned.
Signed-off-by: Vincent Giersch <vincent@giersch.fr>
Instead, provide a variant of instrumentedResponseWriter that does not
implement CloseNotifier, and use that when necessary. In
copyFullPayload, log instead of panicing when we encounter something
that doesn't implement CloseNotifier.
This is more complicated than I'd like, but it's necessary because
instrumentedResponseWriter must not embed CloseNotifier unless there's
really a CloseNotifier to embed.
Signed-off-by: Aaron Lehmann <aaron.lehmann@docker.com>
with a new `proxy` section in the configuration file.
Create a new registry type which delegates storage to a proxyBlobStore
and proxyManifestStore. These stores will pull through data if not present
locally. proxyBlobStore takes care not to write duplicate data to disk.
Add a scheduler to cleanup expired content. The scheduler runs as a background
goroutine. When a blob or manifest is pulled through from the remote registry,
an entry is added to the scheduler with a TTL. When the TTL expires the
scheduler calls a pre-specified function to remove the fetched resource.
Add token authentication to the registry middleware. Get a token at startup
and preload the credential store with the username and password supplied in the
config file.
Allow resumable digest functionality to be disabled at runtime and disable
it when the registry is a pull through cache.
Signed-off-by: Richard Scothern <richard.scothern@gmail.com>
registry/storage/blob_test.go:149: arg d for printf verb %s of wrong type: github.com/docker/distribution.Descriptor
Signed-off-by: Doug Davis <dug@us.ibm.com>
The response code isn't actually sent to the client, because the
connection has already closed by this point. But it causes the status
code to appear as 499 in the logs instead of 0.
Signed-off-by: Aaron Lehmann <aaron.lehmann@docker.com>
When a client disconnects without completing a HTTP request, we were
attempting to process the partial request, which usually leads to a 400
error. These errors can pollute the logs and make it more difficult to
track down real bugs.
This change uses CloseNotifier to detect disconnects. In combination
with checking Content-Length, we can detect a disconnect before sending
the full payload, and avoid logging a 400 error.
This logic is only applied to PUT, POST, and PATCH endpoints, as these
are the places where disconnects during a request are most likely to
happen.
Signed-off-by: Aaron Lehmann <aaron.lehmann@docker.com>
includes the http response.
When debugging non-successful registry requests this will place
the error details and http status fields in the same log line
giving easier visibility to what error occured in the request.
Signed-off-by: Richard Scothern <richard.scothern@gmail.com>
Log a warning if the registry generates its own secret.
Update configuration doc, and remove the default secret from the
development config file.
Signed-off-by: Aaron Lehmann <aaron.lehmann@docker.com>
This passed in the #744 before merge, but apparently the test changed
since the PR was created in ways that led to a new failures.
Signed-off-by: Aaron Lehmann <aaron.lehmann@docker.com>
Storage drivers can implement a method called URLFor which can return a direct
url for a given path. The functionality allows the registry to direct clients
to download content directly from the backend storage. This is commonly used
with s3 and cloudfront. Under certain conditions, such as when the registry is
not local to the backend, these redirects can hurt performance and waste
incoming bandwidth on pulls. This feature addition allows one to disable this
feature, if required.
Signed-off-by: Stephen J Day <stephen.day@docker.com>
Conflicts:
configuration/configuration.go
registry/handlers/app.go
registry/storage/catalog_test.go
registry/storage/manifeststore_test.go
registry/storage/registry.go
Implement the delete API by implementing soft delete for layers
and blobs by removing link files and updating the blob descriptor
cache. Deletion is configurable - if it is disabled API calls
will return an unsupported error.
We invalidate the blob descriptor cache by changing the linkedBlobStore's
blobStatter to a blobDescriptorService and naming it blobAccessController.
Delete() is added throughout the relevant API to support this functionality.
Signed-off-by: Richard Scothern <richard.scothern@gmail.com>
Because the logger was incorrectly replaced while adding hooks, log output did
not include the version and instance ids. The main issue was the the
logrus.Entry was replaced with the logger, which included no context. Replacing
the logger on the context is not necessary when configuring hooks since we are
configuring the contexts logger directly.
Signed-off-by: Stephen J Day <stephen.day@docker.com>
This removes the erroneous http.Handler interface in favor a simple SetHeaders
method that only operattes on the response. Several unnecessary uses of pointer
types were also fixed up.
Signed-off-by: Stephen J Day <stephen.day@docker.com>
Some missing descriptions and error code for tags pagination was cleaned up to
ensure clarity. Specifically, we ensure the request variations are named and
the proper error codes are included.
Signed-off-by: Stephen J Day <stephen.day@docker.com>
The way Repositories() was initially called was somewhat different than
other parts of the client bindings because there was no way to instantiate a
Namespace. This change implements a NewRegistry() function which changes
it so that Repositories() can be called the way one would expect.
It doesn't implement any of the other functions of Namespaces.
Signed-off-by: Patrick Devine <patrick.devine@docker.com>
This change removes the Catalog Service and replaces it with a more
simplistic Repositories() method for obtaining a catalog of all
repositories. The Repositories method takes a pre-allocated slice
and fills it up to the size of the slice and returns the amount
filled. The catalog is returned lexicographically and will start
being filled from the last entry passed to Repositories(). If there
are no more entries to fill, io.EOF will be returned.
Signed-off-by: Patrick Devine <patrick.devine@docker.com>
Conflicts:
registry/client/repository.go
registry/handlers/api_test.go
This change adds a basic catalog endpoint to the API, which returns a list,
or partial list, of all of the repositories contained in the registry. Calls
to this endpoint are somewhat expensive, as every call requires walking a
large part of the registry.
Instead, to maintain a list of repositories, you would first call the catalog
endpoint to get an initial list, and then use the events API to maintain
any future repositories.
Signed-off-by: Patrick Devine <patrick.devine@docker.com>
Move the specification to use a Link header, rather than a "next" entry in the
json results. This prevents requiring clients from parsing the request body to
issue the next request. It also ensures that the returned response body does
not change in between requests.
The ordering of the specification has been slightly tweaked, as well. Listing
image tags has been moved after the catalog specification. Tag pagination now
heavily references catalog pagination.
Signed-off-by: Stephen J Day <stephen.day@docker.com>
This contains a proposal for a catalog API, provided access to the internal
contents of a registry instance. The API endpoint is prefixed with an
underscore, which is illegal in images names, to prevent collisions with
repositories names. To avoid issues with large result sets, a paginated version
of the API is proposed. We make an addition to the tags API to support
pagination to ensure the specification is conistent.
Signed-off-by: Stephen J Day <stephen.day@docker.com>
After consideration, we've changed the main descriptor field name to for number
of bytes to "size" to match convention. While this may be a subjective
argument, commonly we refer to files by their "size" rather than their
"length". This will match other conventions, like `(FileInfo).Size()` and
methods on `io.SizeReaderAt`. Under more broad analysis, this argument doesn't
necessarily hold up. If anything, "size" is shorter than "length".
Signed-off-by: Stephen J Day <stephen.day@docker.com>
An error level log is already produced within app.authorized() if an
actual unexpected error occurs during authorization, so this warning
level log remains for auditability purposes, but should not be
considered an error condition.
Addresses #704
Signed-off-by: Brian Bland <brian.bland@docker.com>
This changeset provides a common http handler for serving errcodes. This should
unify http responses across webservices in the face of errors.
Several type assertions have been added, as well, to ensure the error interface
is implemented.
Signed-off-by: Stephen J Day <stephen.day@docker.com>
Add a functional argument to pass a digest to (ManifestService).GetByTag().
If the digest matches an empty manifest and nil error are returned.
See 1bc740b0d5 for server implementation.
Signed-off-by: Richard Scothern <richard.scothern@gmail.com>
The main goal of this changeset is to allow repository name components to
consist of a single character. The number of components allowed and the slash
separation requirements have also been clarified.
To go along with this simplification, errant constants and unneeded error types
have been removed.
Signed-off-by: Stephen J Day <stephen.day@docker.com>
Challenger manager interface is used to handle getting authorization challenges from an endpoint as well as extracting challenges from responses.
Signed-off-by: Derek McGowan <derek@mcgstyle.net> (github: dmcgowan)
Replace ping logic with individual functions to extract API version and authorization challenges. The response from a ping operation can be passed into these function. If an error occurs in parsing, the version or challenge will not be used. Sending the ping request is the responsibility of the caller.
APIVersion has been converted from a string to a structure type. A parse function was added to convert from string to the structure type.
Signed-off-by: Derek McGowan <derek@mcgstyle.net> (github: dmcgowan)
Move client auth into a separate package.
Separate ping from the authorizer and export Challenges type.
Signed-off-by: Derek McGowan <derek@mcgstyle.net> (github: dmcgowan)
Timeouts should not be a discrete period of time, because they end
up being arbitrary and may be difficult to gauge correctly against
very large Docker layers. Rather, timeouts should be set at the
transport level using the SetDeadline attribute on a net.Conn
object.
Signed-off-by: Jon Poler <jonathan.poler@apcera.com>
This removes documentation and code related to IPC based storage driver
plugins. The existence of this functionality was an original feature goal but
is now not maintained and actively confusing incoming contributions. We will
likely explore some driver plugin mechanism in the future but we don't need
this laying around in the meantime.
Signed-off-by: Stephen J Day <stephen.day@docker.com>
See: 3ea67df373/registry/handlers/app.go (L498)
Per the comment on line 498, this moves the logic of setting the http
status code into the serveJSON func, leaving the auth.Challenge.ServeHTTP()
func to just set the auth challenge header.
Signed-off-by: Doug Davis <dug@us.ibm.com>
This ensures that rados is not required when building the registry. This was
slightly tricky in that when the flags were applied, the rados package was
completely missing. This led to a problem where rados was basically unlistable
and untestable as a package. This was fixed by simply adding a doc.go file that
is included whether rados is built or not.
Signed-off-by: Stephen J Day <stephen.day@docker.com>
This change refreshes the updated version of Azure SDK
for Go that has the latest changes.
I manually vendored the new SDK (github.com/Azure/azure-sdk-for-go)
and I removed `management/` `core/` packages manually simply because
they're not used here and they have a fork of `net/http` and `crypto/tls`
for a particular reason. It was introducing a 44k SLOC change otherwise...
This also undoes the `include_azure` flag (actually Steven removed the
driver from imports but forgot to add the build flag apparently, so the
flag wasn't really including azure. 😄 ). This also must be obsolete
now.
Fixes#620, #175.
Signed-off-by: Ahmet Alp Balkan <ahmetalpbalkan@gmail.com>
After consideration, the basic authentication implementation has been
simplified to only support bcrypt entries in an htpasswd file. This greatly
increases the security of the implementation by reducing the possibility of
timing attacks and other problems trying to detect the password hash type.
Also, the htpasswd file is only parsed at startup, ensuring that the file can
be edited and not effect ongoing requests. Newly added passwords take effect on
restart. Subsequently, password hash entries are now stored in a map.
Test cases have been modified accordingly.
Signed-off-by: Stephen J Day <stephen.day@docker.com>
This change refactors the basic authentication implementation to better follow
Go coding standards. Many types are no longer exported. The parser is now a
separate function from the authentication code. The standard functions
(*http.Request).BasicAuth/SetBasicAuth are now used where appropriate.
Signed-off-by: Stephen J Day <stephen.day@docker.com>
In the request parameters lists `tag` was used instead of
`reference` present in the HTTP requests paths
Signed-off-by: Vincent Giersch <vincent.giersch@ovh.net>
This PR is for issue of "email after registry webapp panic" #41, improving my
previous design (closed).
It use self setting up hooks, to catch panic in web application.
And, send email in hooks handle directly, to no use new http server and
handler.
Signed-off-by: xiekeyang <keyangxie@126.com>
To make the definition of supported digests more clear, we have refactored the
digest package to have a special Algorithm type. This represents the digest's
prefix and we associated various supported hash implementations through
function calls.
Signed-off-by: Stephen J Day <stephen.day@docker.com>
The change relies on a refactor of the upstream resumable sha256/sha512 package
that opts to register implementations with the standard library. This allows
the resumable support to be detected where it matters, avoiding unnecessary and
complex code. It also ensures that consumers of the digest package don't need
to depend on the forked sha implementations.
We also get an optimization with this change. If the size of data written to a
digester is the same as the file size, we check to see if the digest has been
verified. This works if the blob is written and committed in a single request.
Signed-off-by: Stephen J Day <stephen.day@docker.com>
Ensure that clients can use the blob descriptor cache provider without needing
the redis package.
Signed-off-by: Stephen J Day <stephen.day@docker.com>
Changes behavior so ping doesn't happen if /v2/ is anywhere in a request path, but instead only at the beginning. This fixes attempts to ping on redirected URLs.
Signed-off-by: Derek McGowan <derek@mcgstyle.net> (github: dmcgowan)
The transport package no longer requires importing distribution for the ReadSeekCloser, instead declares its own.
Added comments on the Authenication handler in session.
Added todo on http seek reader to highlight its lack of belonging to the client transport.
Signed-off-by: Derek McGowan <derek@mcgstyle.net> (github: dmcgowan)
Add check for unauthorized error code and explicitly set the error code if the content could not be parsed.
Updated repository test for unauthorized tests and nit feedback.
Signed-off-by: Derek McGowan <derek@mcgstyle.net> (github: dmcgowan)
Each type no longer requires holding a reference to repository.
Added implementation for signatures get.
Signed-off-by: Derek McGowan <derek@mcgstyle.net> (github: dmcgowan)
Update comments and TODOs
Fix switch style
Updated parse http response to take in reader
Add Cancel implementation
Update blobstore variable name
Signed-off-by: Derek McGowan <derek@mcgstyle.net> (github: dmcgowan)
This driver implements the storagedriver.StorageDriver interface and
uses Ceph Object Storage as storage backend.
Since RADOS is an object storage and no hierarchy notion, the
following convention is used to keep the filesystem notions stored in
this backend:
* All the objects data are stored with opaque UUID names prefixed
(e.g. "blob:d3d232ff-ab3a-4046-9ab7-930228d4c164).
* All the hierarchy information are stored in rados omaps, where the
omap object identifier is the virtual directory name, the keys in
a specific are the relative filenames and the values the blob
object identifier (or empty value for a sub directory).
e.g. For the following hierarchy:
/directory1
/directory1/object1
/directory1/object2
/directory1/directory2/object3
The omap "/directory1" will contains the following key / values:
- "object1" "blob:d3d232ff-ab3a-4046-9ab7-930228d4c164"
- "object2" "blob:db2e359d-4af0-4bfb-ba1d-d2fd029866a0"
- "directory2" ""
The omap "/directory1/directory2" will contains:
- "object3" "blob:9ae2371c-81fc-4945-80ac-8bf7f566a5d9"
* The MOVE is implemented by changing the reference to a specific
blob in its parent virtual directory omap.
This driver stripes rados objects to a fixed size (e.g. 4M). The idea
is to keep small objects (as done by RBD on the top of RADOS) that
will be easily synchronized accross OSDs. The information of the
original object (i.e total size of the chunks) is stored as a Xattr
in the first chunk object.
Signed-off-by: Vincent Giersch <vincent.giersch@ovh.net>
Repository creation now just takes in an http.RoundTripper. Authenticated requests or requests which require additional headers should use the NewTransport function along with a request modifier (such an an authentication handler).
Signed-off-by: Derek McGowan <derek@mcgstyle.net> (github: dmcgowan)
Wrapping the reader in a NopCloser is necessary to prevent the http library from closing the input reader.
Signed-off-by: Derek McGowan <derek@mcgstyle.net> (github: dmcgowan)
Refactory authorizer to take a set of authentication handlers for different authentication schemes returned by an unauthorized HTTP requst.
Signed-off-by: Derek McGowan <derek@mcgstyle.net> (github: dmcgowan)
Adds functionality to create a Repository client which connects to a remote endpoint.
Signed-off-by: Derek McGowan <derek@mcgstyle.net> (github: dmcgowan)
This PR refactors the blob service API to be oriented around blob descriptors.
Identified by digests, blobs become an abstract entity that can be read and
written using a descriptor as a handle. This allows blobs to take many forms,
such as a ReadSeekCloser or a simple byte buffer, allowing blob oriented
operations to better integrate with blob agnostic APIs (such as the `io`
package). The error definitions are now better organized to reflect conditions
that can only be seen when interacting with the blob API.
The main benefit of this is to separate the much smaller metadata from large
file storage. Many benefits also follow from this. Reading and writing has
been separated into discrete services. Backend implementation is also
simplified, by reducing the amount of metadata that needs to be picked up to
simply serve a read. This also improves cacheability.
"Opening" a blob simply consists of an access check (Stat) and a path
calculation. Caching is greatly simplified and we've made the mapping of
provisional to canonical hashes a first-class concept. BlobDescriptorService
and BlobProvider can be combined in different ways to achieve varying effects.
Recommend Review Approach
-------------------------
This is a very large patch. While apologies are in order, we are getting a
considerable amount of refactoring. Most changes follow from the changes to
the root package (distribution), so start there. From there, the main changes
are in storage. Looking at (*repository).Blobs will help to understand the how
the linkedBlobStore is wired. One can explore the internals within and also
branch out into understanding the changes to the caching layer. Following the
descriptions below will also help to guide you.
To reduce the chances for regressions, it was critical that major changes to
unit tests were avoided. Where possible, they are left untouched and where
not, the spirit is hopefully captured. Pay particular attention to where
behavior may have changed.
Storage
-------
The primary changes to the `storage` package, other than the interface
updates, were to merge the layerstore and blobstore. Blob access is now
layered even further. The first layer, blobStore, exposes a global
`BlobStatter` and `BlobProvider`. Operations here provide a fast path for most
read operations that don't take access control into account. The
`linkedBlobStore` layers on top of the `blobStore`, providing repository-
scoped blob link management in the backend. The `linkedBlobStore` implements
the full `BlobStore` suite, providing access-controlled, repository-local blob
writers. The abstraction between the two is slightly broken in that
`linkedBlobStore` is the only channel under which one can write into the global
blob store. The `linkedBlobStore` also provides flexibility in that it can act
over different link sets depending on configuration. This allows us to use the
same code for signature links, manifest links and blob links. Eventually, we
will fully consolidate this storage.
The improved cache flow comes from the `linkedBlobStatter` component
of `linkedBlobStore`. Using a `cachedBlobStatter`, these combine together to
provide a simple cache hierarchy that should streamline access checks on read
and write operations, or at least provide a single path to optimize. The
metrics have been changed in a slightly incompatible way since the former
operations, Fetch and Exists, are no longer relevant.
The fileWriter and fileReader have been slightly modified to support the rest
of the changes. The most interesting is the removal of the `Stat` call from
`newFileReader`. This was the source of unnecessary round trips that were only
present to look up the size of the resulting reader. Now, one must simply pass
in the size, requiring the caller to decide whether or not the `Stat` call is
appropriate. In several cases, it turned out the caller already had the size
already. The `WriterAt` implementation has been removed from `fileWriter`,
since it is no longer required for `BlobWriter`, reducing the number of paths
which writes may take.
Cache
-----
Unfortunately, the `cache` package required a near full rewrite. It was pretty
mechanical in that the cache is oriented around the `BlobDescriptorService`
slightly modified to include the ability to set the values for individual
digests. While the implementation is oriented towards caching, it can act as a
primary store. Provisions are in place to have repository local metadata, in
addition to global metadata. Fallback is implemented as a part of the storage
package to maintain this flexibility.
One unfortunate side-effect is that caching is now repository-scoped, rather
than global. This should have little effect on performance but may increase
memory usage.
Handlers
--------
The `handlers` package has been updated to leverage the new API. For the most
part, the changes are superficial or mechanical based on the API changes. This
did expose a bug in the handling of provisional vs canonical digests that was
fixed in the unit tests.
Configuration
-------------
One user-facing change has been made to the configuration and is updated in
the associated documentation. The `layerinfo` cache parameter has been
deprecated by the `blobdescriptor` cache parameter. Both are equivalent and
configuration files should be backward compatible.
Notifications
-------------
Changes the `notification` package are simply to support the interface
changes.
Context
-------
A small change has been made to the tracing log-level. Traces have been moved
from "info" to "debug" level to reduce output when not needed.
Signed-off-by: Stephen J Day <stephen.day@docker.com>
- Set an Etag header
- Check If-None-Match and respond appropriately
- Set a Cache-Control header with a default of 1 week
Signed-off-by: Richard Scothern <richard.scothern@gmail.com>
Allow to use a unix socket as a listener.
To specify an endpoint type we use an optional configuration
field 'net', as there's no way to distinguish a relative
socket path from a hostname.
Signed-off-by: Anton Tiurin <noxiouz@yandex.ru>
- Ensures new uploads and resumed upload statuses always return an offset of 0. This allows future clients which support resumable upload to not attempt resumable upload on this version which does not support it.
- Add PATCH support for streaming data on upload.
- Add messaging to specification that PATCH with content range is currently not supported.
- Update PUT blob to only support full data or no data, no more last chunk messaging as it was not supported.
closes#470
Signed-off-by: Derek McGowan <derek@mcgstyle.net> (github: dmcgowan)
This deals with a memory leak, caused by goroutines, experienced when using the
s3 driver. Unfortunately, this section of the code leaks goroutines like a
sieve. There is probably some refactoring that could be done to avoid this but
instead, we have a done channel that will cause waiting goroutines to exit.
Signed-off-by: Stephen J Day <stephen.day@docker.com>
- Change driver interface to take a context as its first argument
- Make newFileReader take a context as its first argument
- Make newFileWriter take a context as its first argument
- Make blobstore exists and delete take a context as a first argument
- Pass the layerreader's context to the storage layer
- Pass the app's context to purgeuploads
- Store the app's context into the blobstore (was previously null)
- Pass the trace'd context to the storage drivers
Signed-off-by: Richard Scothern <richard.scothern@gmail.com>
According to the Apache mod_proxy docs, X-Forwarded-Host can be a
comma-separated list of hosts, to which each proxy appends the requested
host. We want to grab only the first from this comma-separated list
to get the original requested Host when building URLs.
Docker-DCO-1.1-Signed-off-by: Josh Hawn <josh.hawn@docker.com> (github: jlhawn)
This adds a missing return statement. It is not strictly needed since if the
io.Copy fails, the Finish operation will fail. Currently, the client reports
both errors where this new code will correctly only report the io.Copy error.
Signed-off-by: Stephen J Day <stephen.day@docker.com>
The code using values from the yaml package wasn't careful enought with the
possible incoming types. Turns out, it is just an int but we've made this
section somewhat bulletproof in case that package changes the behavior.
This code likely never worked. The configuration system should be decoupled
from the object instantiation.
Signed-off-by: Stephen J Day <stephen.day@docker.com>
Rather than accept the resulting of a layer validation, we retry up to three
times, backing off 100ms after each try. The thought is that we allow s3 files
to make their way into the correct location increasing the liklihood the
verification can proceed, if possible.
Signed-off-by: Stephen J Day <stephen.day@docker.com>
When the registry starts a background timer will periodically
scan the upload directories on the file system every 24 hours
and delete any files older than 1 week. An initial jitter
intends to avoid contention on the filesystem where multiple
registries with the same storage driver are started
simultaneously.
Ensure that the status is logged in the context by instantiating before the
request is routed to handlers. While this requires some level of hacking to
acheive, the result is that the context value of "http.request.status" is as
accurate as possible for each request.
Signed-off-by: Stephen J Day <stephen.day@docker.com>
Registry is intended to be used as a repository service than an abstract collection of repositories. Namespace better describes a collection of repositories retrievable by name.
The registry service serves any repository in the global scope.
Signed-off-by: Derek McGowan <derek@mcgstyle.net> (github: dmcgowan)
This moves the instance id out of the app so that it is associated with an
instantiation of the runtime. The instance id is stored on the background
context. This allows allow contexts using the main background context to
include an instance id for log messages. It also simplifies the application
slightly.
Signed-off-by: Stephen J Day <stephen.day@docker.com>
The original implementation wrote to different locations in a shared slice.
While this is theoretically okay, we end up thrashing the cpu cache since
multiple slice members may be on the same cache line. So, even though each
thread has its own memory location, there may be contention over the cache
line. This changes the code to aggregate to a slice in a single goroutine.
In reality, this change likely won't have any performance impact. The theory
proposed above hasn't really even been tested. Either way, we can consider it
and possibly go forward.
Signed-off-by: Stephen J Day <stephen.day@docker.com>
Rather than enforce lowercase paths for all drivers, support for
case-sensitivity has been deferred to the driver. There are a few caveats to
this approach:
1. There are possible security implications for tags that only differ in their
case. For instance, a tag "A" may be equivalent to tag "a" on certain file
system backends.
2. All system paths should not use case-sensitive identifiers where possible.
This might be problematic in a blob store that uses case-sensitive ids. For
now, since digest hex ids are all case-insensitive, this will not be an issue.
The recommend workaround is to not run the registry on a case-insensitive
filesystem driver in security sensitive applications.
Signed-off-by: Stephen J Day <stephen.day@docker.com>
To avoid compounded round trips leading to slow retrieval of manifests with a
large number of signatures, the fetch of signatures has been parallelized. This
simply spawns a goroutine for each path, coordinated with a sync.WaitGroup.
Signed-off-by: Stephen J Day <stephen.day@docker.com>
For consistency with other systems, the redis and caching monitoring data has
been moved under the "registry" section in expvar. This ensures the entire
registry state is kept to a single section.
Signed-off-by: Stephen J Day <stephen.day@docker.com>
This allows one to better control the usage of the cache and turn it off
completely. The storage configuration module was modified to allow parameters
to be passed to just the storage implementation, rather than to the driver.
Signed-off-by: Stephen J Day <stephen.day@docker.com>
This changeset integrates the layer info cache with the registry webapp and
storage backend. The main benefit is to cache immutable layer meta data,
reducing backend roundtrips. The cache can be configured to use either redis or
an inmemory cache.
This provides massive performance benefits for HEAD http checks on layer blobs
and manifest verification.
Signed-off-by: Stephen J Day <stephen.day@docker.com>
This changeset defines the interface for layer info caches. Layer info caches
speed up access to layer meta data accessed in storage driver backends. The
two main operations are tests for repository membership and resolving path and
size information for backend blobs.
Two implementations are available. The main implementation leverages redis to
store layer info. An alternative implementation simply caches layer info in
maps, which should speed up resolution for less sophisticated implementations.
Signed-off-by: Stephen J Day <stephen.day@docker.com>
By using a resumable digester and storing the state of upload digests between
subsequent upload chunks, finalizing an upload no longer requires reading back
all of the uploaded data to verify the client's expected digest.
Docker-DCO-1.1-Signed-off-by: Josh Hawn <josh.hawn@docker.com> (github: jlhawn)
This chnage prevents a crash when moving from a non-existent directory that has
a file as a parent. To prevent this, we simply check that the node is a
directory and throws an error if it is not.
Signed-off-by: Stephen J Day <stephen.day@docker.com>
This change adds strong validation for the uuid variable for v2 routes. This is
a minor specification change but is okay since the uuid field is controlled by
the server. The character set is restricted to avoid path traversal, allowing
for alphanumeric values and urlsafe base64 encoding.
This change has no effect on client implementations.
Signed-off-by: Stephen J Day <stephen.day@docker.com>
Redis has been integrated with the web application for use with various
services. The configuraiton exposes connection details, timeouts and pool
parameters. Documentation has been updated accordingly.
A few convenience methods have been added to the context package to get loggers
with certain fields, exposing some missing functionality from logrus.
Signed-off-by: Stephen J Day <stephen.day@docker.com>
This enables Azure storage driver to be used with non-default
cloud endpoints like Azure China or Azure Government that does
not use `.blob.core.windows.net` FQDN suffix.
Signed-off-by: Ahmet Alp Balkan <ahmetalpbalkan@gmail.com>
Convert middleware in the config to be a map of type->[]Middleware
Add support for registry & repository middleware.
Some naming updates as well.
Signed-off-by: Andy Goldstein <agoldste@redhat.com>
middleware concept.
This also breaks the dependency the storage package had on goamz
Signed-off-by: David Lawrence <david.lawrence@docker.com> (github: endophage)
This changeset implements immutable manifest references via the HTTP API. Most
of the changes follow from modifications to ManifestService. Once updates were
made across the repo to implement these changes, the http handlers were change
accordingly. The new methods on ManifestService will be broken out into a
tagging service in a later PR.
Unfortunately, due to complexities around managing the manifest tag index in an
eventually consistent manner, direct deletes of manifests have been disabled.
Signed-off-by: Stephen J Day <stephen.day@docker.com>
Manifests are now fetched by a field called "reference", which may be a tag or
a digest. When using digests to reference a manifest, the data is immutable.
The routes and specification have been updated to allow this.
There are a few caveats to this approach:
1. It may be problematic to rely on data format to differentiate between a tag
and a digest. Currently, they are disjoint but there may modifications on
either side that break this guarantee.
2. The caching characteristics of returned content are very different for
digest versus tag-based references. Digest urls can be cached forever while tag
urls cannot.
Both of these are minimal caveats that we can live with in the future.
Signed-off-by: Stephen J Day <stephen.day@docker.com>
