When images are pulled by digest only (not by tag), the API endpoint
was returning 0 for manifest_pull_count because it ignored manifest_stats
when tag_stats was None.
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-authored-by: Claude <noreply@anthropic.com>
Implements global read-only superuser permissions for v1 endpoints, adjusts superuser write checks, and updates app token listing and detail endpoints; includes comprehensive tests.
---------
Co-authored-by: Claude <noreply@anthropic.com>
* v2: Disallow push of manifests with negative layer size (PROJQUAY-8560)
Under certain conditions, clients may create a manifest (OCI or Docker v2) that contains negative layer sizes. Our current validation schema does not take that corner case into account, it only checks if the manifest is properly formatted or not. With this change, Quay will reject manifests that have negative layer sizes and raise a `400` with a proper exception. An example can be seen here:
~~~
gunicorn-registry stdout | 2025-02-10 22:34:54,930 [377] [ERROR] [endpoints.v2.manifest] failed to parse manifest when writing by tagname
gunicorn-registry stdout | Traceback (most recent call last):
gunicorn-registry stdout | File "/quay-registry/endpoints/v2/manifest.py", line 362, in _parse_manifest
gunicorn-registry stdout | return parse_manifest_from_bytes(
gunicorn-registry stdout | File "/quay-registry/image/shared/schemas.py", line 40, in parse_manifest_from_bytes
gunicorn-registry stdout | return DockerSchema2Manifest(manifest_bytes)
gunicorn-registry stdout | File "/quay-registry/image/docker/schema2/manifest.py", line 172, in __init__
gunicorn-registry stdout | raise MalformedSchema2Manifest("layer size is negative")
gunicorn-registry stdout | image.docker.schema2.manifest.MalformedSchema2Manifest: layer size is negative
gunicorn-registry stdout | 2025-02-10 22:34:54,931 [377] [DEBUG] [endpoints.v2] sending response: b'{"errors":[{"code":"MANIFEST_INVALID","detail":{"message":"failed to parse manifest: layer size is negative"},"message":"manifest invalid"}]}\n
~~~
* Add tests, changed error message on malformed manifest exception
* Fix v2 API test
* Add match expression to pytest to make sure a proper exception is raised
* Add exception for layers with size 0 bytes, fix tests
* Fix indentation on previous changes
* Fix indetation
* storage: Disable pushes on registry (PROJQUAY-6870)
The current read-only option for Quay is not sometimes feasible, since it requires an insert of the service key and other manual config changes. For instance, if you want to just recalculate quota on the registry, but would like to allow all registry operations (including UI) without the possibility of pushes until recalculation is done, setting the whole registry `read-only` cannot be done since it makes the database read only as well.
This PR introduces a new flag called `DISABLE_PUSHES` which allows all registry operations to continue (changing tags, repo editing, robot account creation/deletion, user creation etc.) but will disable pushes of new images to the registry (i.e. backend storage will not change). If a registry already contains the image and a new tag is simply being added, that operation should succeed.
The following message would appear in the logs:
~~~
gunicorn-registry stdout | 2024-03-13 20:19:49,414 [369] [DEBUG] [endpoints.v2] sending response: b'{"errors":[{"code":"METHOD NOT ALLOWED","detail":{},"message":"Pushes to the registry are currently disabled. Please contact the administrator for more information."}]}\n'
gunicorn-registry stdout | 2024-03-13 20:19:49,414 [369] [INFO] [gunicorn.access] 172.17.0.1 - - [13/Mar/2024:20:19:49 +0000] "PUT /v2/ibazulic/mariadb/manifests/sha256:c4694ba424e0259694a5117bbb510d67340051f0bdb7f9fa8033941a2d66e53e HTTP/1.1" 405 169 "-" "skopeo/1.9.3"
nginx stdout | 172.17.0.1 (-) - - [13/Mar/2024:20:19:49 +0000] "PUT /v2/ibazulic/mariadb/manifests/sha256:c4694ba424e0259694a5117bbb510d67340051f0bdb7f9fa8033941a2d66e53e HTTP/1.1" 405 169 "-" "skopeo/1.9.3" (0.002 3813 0.002)
~~~
The flag defaults to `False` (pushes enabled), unless set otherwise.
* Removed constraint on storage replication when pushes are disabled
* Rebase
* Fix isort sorting
* Fix isort sorting #2
* Removed constraint on storage replication when pushes are disabled
* Rebase
* Remove constraint on storage replication worker
* Fix linting on config.py
introduces a check of the OrganizationMemberPermission for pulls
against a proxy org.
if the proxy cache feature is disabled, or the org is not a proxy org
the check is not performed and Quay will behave normally.
this check does not mean pulls will work transparently though -
non-admin users need to be added to a team in the proxy org with the
member role, and default read and write permissions need to be given to
that team so that non-admin users can pull and update the cache for
images they do not own (the user who first pulls an image ends up
owning the repository since that is when the repo gets created).
Currently when Quay is set to automatically create organizations it will check the user table for a matching username with organization set to true. This causes conflicts when a user already exists with the same username, where the check will fail and Quay will attempt to create an organization with the same name as the user. This change checks for only a matching username.
introduces the possibility to pull images from external registries
through Quay, storing them locally for faster subsequent pulls.
Closes PROJQUAY-3030 and PROJQUAY-3033
New config CREATE_PRIVATE_REPO_ON_PUSH to control global default for new repositories created via push. Defaults to True, meaning the repo will be created with "private" visibility. If set to False, new repos created by push will be created with "public" visibility.
* Change verbs to use a DerivedStorageForManifest table instead of DerivedStorageForImage
This allows us to deprecate the DerivedStorageForImage table.
Fixes https://issues.redhat.com/browse/PROJQUAY-519
* Change uploaded blob tracking to use its own table and deprecate
RepositoryTag
* Start recording the compressed layers size and config media type on the
manifest row in the database
NOTE: This change includes a database migration which will *lock* the
manifest table
* Change tag API to return the layers size from the manifest
* Remove unused code
* Add new config_media_type field to OCI types
* Fix secscan V2 test for us no longer writing temp images
* Remove unused uploading field
* Switch registry model to use synthetic legacy images
Legacy images are now (with exception of the V2 security model) read from the *manifest* and sythensized in memory. The legacy image IDs are generated realtime based on the hashids library. This change also further deprecates a bunch of our Image APIs, reducing them to only returning the image IDs, and emptying out the remaining metadata (to avoid the requirement of us loading the information for the manifest from storage).
This has been tested with our full clients test suite with success.
* Add a backfill worker for manifest layers compressed sizes
* Change image tracks into manifest tracks now that we no longer have
manifest-less tags
* Add back in the missing method
* Add missing joins to reduce extra queries
* Remove unnecessary join when looking up legacy images
* Remove extra hidden filter on tag queries
* Further DB improvements
* Delete all Verbs, as they were deprecated
* Add back missing parameter in manifest data type
* Fix join to return None for the robot if not defined on mirror config
* switch to using secscan_v4_model for all indexing and remove most of secscan_v2_model code
* Add a missing join
* Remove files accidentally re-added due to rebase
* Add back hashids lib
* Rebase fixes
* Fix broken test
* Remove unused GPG signer now that ACI conversion is removed
* Remove duplicated repomirrorworker
* Remove unused notification code for secscan. We'll re-add it once Clair
V4 security notifications are ready to go
* Fix formatting
* Stop writing Image rows when creating manifests
* Stop writing empty layer blobs for manifests
As these blobs are shared, we don't need to write ManifestBlob rows
for them
* Remove further unused code
* Add doc comment to _build_blob_map
* Add unit test for synthetic V1 IDs
* Remove unused import
* Add an invalid value test to synthetic ID decode tests
* Add manifest backfill worker back in
Seems to have been removed at some point
* Add a test for cached active tags
* Rename test_shared to not conflict with another same-named test file
Pytest doesn't like having two test modules with the same name
* Have manifestbackfillworker also copy over the config_media_type if present
Co-authored-by: alecmerdler <alecmerdler@gmail.com>
* Convert all Python2 to Python3 syntax.
* Removes oauth2lib dependency
* Replace mockredis with fakeredis
* byte/str conversions
* Removes nonexisting __nonzero__ in Python3
* Python3 Dockerfile and related
* [PROJQUAY-98] Replace resumablehashlib with rehash
* PROJQUAY-123 - replace gpgme with python3-gpg
* [PROJQUAY-135] Fix unhashable class error
* Update external dependencies for Python 3
- Move github.com/app-registry/appr to github.com/quay/appr
- github.com/coderanger/supervisor-stdout
- github.com/DevTable/container-cloud-config
- Update to latest mockldap with changes applied from coreos/mockldap
- Update dependencies in requirements.txt and requirements-dev.txt
* Default FLOAT_REPR function to str in json encoder and removes keyword assignment
True, False, and str were not keywords in Python2...
* [PROJQUAY-165] Replace package `bencode` with `bencode.py`
- Bencode is not compatible with Python 3.x and is no longer
maintained. Bencode.py appears to be a drop-in replacement/fork
that is compatible with Python 3.
* Make sure monkey.patch is called before anything else (
* Removes anunidecode dependency and replaces it with text_unidecode
* Base64 encode/decode pickle dumps/loads when storing value in DB
Base64 encodes/decodes the serialized values when storing them in the
DB. Also make sure to return a Python3 string instead of a Bytes when
coercing for db, otherwise, Postgres' TEXT field will convert it into
a hex representation when storing the value.
* Implement __hash__ on Digest class
In Python 3, if a class defines __eq__() but not __hash__(), its
instances will not be usable as items in hashable collections (e.g sets).
* Remove basestring check
* Fix expected message in credentials tests
* Fix usage of Cryptography.Fernet for Python3 (#219)
- Specifically, this addresses the issue where Byte<->String
conversions weren't being applied correctly.
* Fix utils
- tar+stream layer format utils
- filelike util
* Fix storage tests
* Fix endpoint tests
* Fix workers tests
* Fix docker's empty layer bytes
* Fix registry tests
* Appr
* Enable CI for Python 3.6
* Skip buildman tests
Skip buildman tests while it's being rewritten to allow ci to pass.
* Install swig for CI
* Update expected exception type in redis validation test
* Fix gpg signing calls
Fix gpg calls for updated gpg wrapper, and add signing tests.
* Convert / to // for Python3 integer division
* WIP: Update buildman to use asyncio instead of trollius.
This dependency is considered deprecated/abandoned and was only
used as an implementation/backport of asyncio on Python 2.x
This is a work in progress, and is included in the PR just to get the
rest of the tests passing. The builder is actually being rewritten.
* Target Python 3.8
* Removes unused files
- Removes unused files that were added accidentally while rebasing
- Small fixes/cleanup
- TODO tasks comments
* Add TODO to verify rehash backward compat with resumablehashlib
* Revert "[PROJQUAY-135] Fix unhashable class error" and implements __hash__ instead.
This reverts commit 735e38e3c1d072bf50ea864bc7e119a55d3a8976.
Instead, defines __hash__ for encryped fields class, using the parent
field's implementation.
* Remove some unused files ad imports
Co-authored-by: Kenny Lee Sin Cheong <kenny.lee@redhat.com>
Co-authored-by: Tom McKay <thomasmckay@redhat.com>
* Remove V3_UPGRADE_MODE
* Remove tag backfill worker and all callers to tag backfill and upgrades
* Change initdb to create all test data via the manifest builder, rather than manually via legacy images
* Convert various code paths to use the registry_model where we previously did not do so
* Convert the GC test suite to build via the manifest builder
* Delete an old, unused tool
* Delete the Pre OCI registry model
* Add additional error handling to the manifest creation code path
* Add additional error handling to the OCI tag creation code path
* Change how we respond to invalid manifest content types to better handle unknowns
* Change legacy secscan test suite to use the registry model
* Change the repo build badge to use the registry model (also fixes a bug)
* Delete now-unused data model code
* Remove old model adjustment code from OCI model
* Mark older data models as deprecated which will prevent new rows from being inserted
* Remove references to old registry test suite from various testing files
* Remove tag backfill worker (again; got re-added during rebase)
* Move all deprecated model checks into a central function
* Make data_migration more Pythonic
* Small requested fixes to Tag module styling
* Have tag backfill worker fail to migrate if there are TagManifest's
Since this backfill should only be called in future releases for empty models, this should catch someone attempting to upgrade from a too-old version
* Remove labelbackfillworker as it is no longer needed
* Remove unused invalid import
* Reimplement the tag test for the remaining method used
