Release Notes¶
WaiverDB 1.4.0¶
Released 10 November 2021
On an authentication error, waiverdb-cli will now print the actual error message instead of a JSONDecodeError exception.
Documentation has been updated and moved to: https://waiverdb.readthedocs.io
WaiverDB 1.3.0¶
Released 26 April 2021
A simpler permission configuration can be now used. New option
PERMISSIONS
uses glob expressions instead of a regular expression to verify the user is allowed to waive a test case. The configuration can be listed withGET /api/v1.0/permissions
. Additional GET parametertestcase
can be used to filter the list by matching test case.PERMISSION_MAPPING
is deprecated but still works if the new option is not defined.Fixes possible disconnecting issue with older stomp.py library (https://github.com/jasonrbriggs/stomp.py/issues/323).
WaiverDB 1.2.0¶
Released 7 January 2021
scenario
field is now supported for both REST API and CLI
WaiverDB 1.1.7¶
Released 18 November 2020
Support for multiple LDAP queries was added
Calling
start
for thestomp.py
connection was removed as it was deprecatedscenario
field was added to the database
WaiverDB 1.1.6¶
Released 10 March 2020
Version 1.1.5 was bumped due to error during the release
WaiverDB 1.1.5¶
Released 9 March 2020
STOMP connection will always disconnect regardless of errors
Added
docker-compose
support
WaiverDB 1.1.4¶
Released 11 November 2019
New
/config
API endpoint to expose the application configuration, i.e. PERMISSION_MAPPING and SUPERUSERS.Retry sending STOMP message after a delay
Revert allow overriding krb_principal option for waiverdb-cli
WaiverDB 1.1.3¶
Released 5 August 2019
Allow overriding krb_principal option for waiverdb-cli
Code optimizations and improvements
WaiverDB 1.1.2¶
Released 2 July 2019
Disable sphinxcontrib-issuetracker integration: this extension appears to no longer be maintained. The following issue prevents adopting a newer version of Sphinx: https://github.com/ignatenkobrain/sphinxcontrib-issuetracker/issues/23
WaiverDB 1.1.1¶
Released 20 May 2019
Use flask-cors library to support CORS headers: moves the CORS header support to the library.
WaiverDB 1.1.0¶
Released 15 May 2019
Restrict waiver creation based on users/groups and testcase: introduce access control based on the users/groups and the testcase. Groups need to be defined in LDAP. New configuration is required to enable this feature:
PERMISSION_MAPPING
: dictionary with keys as regex applied to test cases and values as dictionaries with “users” and “groups” allowed to submit waivers for that matching test case. If not specified, the feature is not enabled.LDAP_HOST
andLDAP_BASE
: required to query the LDAP system. Required ifPERMISSION_MAPPING
is defined.
Add
proxied_by
in the CLI: in the API is possible to define a username on whose behalf the caller is proxying the submittion of a waiver. This change provides this possibility also in the CLI. This is updated also to reflect the recent changes regarding the access control.Allow optional trailing slash for about endpoint: accessing “api/v1.0/about/” won’t give 404 anymore.
WaiverDB 1.0.0¶
Replace fedmsg with fedora-messaging
Don’t validate the CA of the server when downloading a custom CA when the container starts
WaiverDB 0.14.1¶
Treat “brew-build” subject type as an alias for “koji_build”: Greenwave expects only values of “koji_build” since that’s what the policies apply to.
WaiverDB 0.14.0¶
Fix incorrect splitting of Python files into subpackages: Since the -common subpackage installs files in %{python3_sitelib}/%{name}, it should also own that directory instead of the main package. Also, exclude files that are in subpackages from the main package. Finally, byte-compiled files were not specified correctly for Python 3. This was hidden in the main package glob, instead of being in the subpackages.
Improve authentication error response: when using Kerberos authentication it is necessary to configure “dns_canonicalize_hostname” inside the Kerberos configuration file otherwise the user will get an error response that is a bit ambiguous. There’s no way to detect if that was the problem or it is just a generic authentication error. But we can provide an hint to the user, and also advise to check the doc.
Introduce a /metrics endpoint to the API for monitoring reasons.
WaiverDB 0.13.0¶
Release 11 January 2019.
Stop validating subject types against a hard-coded list. Since Greenwave now supports arbitrary subject types, this list of valid subject types no longer needs to be maintained.
WaiverDB 0.12.0¶
Released 8 November 2018.
Invalid
subject
values are handled during database migration.The waiverdb-cli utility accepts new options
--subject-identifier
and--subject-type
which deprecate--subject
option.python-requests-gssapi is now a
requires
andbuildrequires
dependency.Locked DB scenario checked in
GET /healthcheck
API endpoint.
WaiverDB 0.11.0¶
Released 3 July 2018.
Waivers now have two new attributes,
subject_type
andsubject_identifier
, to identify the subject of the waiver (that is, the particular software artifact that the waiver is about). These new attributes replace thesubject
attribute which is now deprecated.The
subject
attribute previously accepted any arbitrary key-values, but in practice thesubject
had to conform to one of several recognized structures in order to be usable with Greenwave. This has now been made explicit with thesubject_type
attribute. See Subject types in the Greenwave documentation for a list of possible subject types and the meaning of their corresponding identifiers. See Greenwave issue 126 for more background about this change.For backwards compatibility the
subject
attribute is still included when fetching waivers, and accepted when creating waivers. However if you create a new waiver using the deprecatedsubject
attribute, its structure must match one of the recognized subject types, otherwise the request will fail with 400 status code. In this release we have implemented support for all known subject types in the wild.New endpoint
POST /api/v1.0/waivers/+filtered
deprecatesPOST /api/v1.0/waivers/+by-subjects-and-testcases
. This allows posting an arbitrary set of filter criteria, instead of the using a complicated and limited API.The waiverdb-cli utility will now guess a suitable default value for the
--product-version
option in many common cases, in order to make it easier to submit waivers (#111). Automated scripts should prefer to explicitly pass--product-version
in case the guessing logic does not work in all cases.Previously, when you requested a list of waivers, WaiverDB would consider waivers from unrelated users and product versions to obsolete each other (#137). The API now correctly returns the most recent waiver from each user, and for each product version.
The documentation now includes a section describing how end users can submit waivers using the command-line interface (see User Guide, #149).
New man page available for
waiverdb-client.conf(5)
(see waiverdb-cli Configuration).
WaiverDB 0.10.0¶
Released 10 May 2018.
Comment is now explicitly required when creating waivers (both in API and CLI).
Multiple waivers can now be created with single POST request (#98). To create multiple waivers, POST list to “waivers/” instead of single waiver.
When creating a waiver by referring to a result ID, WaiverDB now accepts results with
'type': 'brew-build'
as an alias for'koji_build'
.Messaging can be disabled is settings with
MESSAGE_PUBLISHER = None
.The
KERBEROS_HTTP_HOST
setting in the server configuration is now ignored. This setting is no longer needed because GSSAPI will automatically find a key in the Kerberos keytab matching the service principal in the client request.New man pages are available for
waiverdb-cli(1)
andwaiverdb(7)
(REST API).Changed error message for bad
since
value. E.g.api/v1.0/waivers/?since=123
results in HTTP 400 with message{"message": {"since": "time data '123' does not match format '%Y-%m-%dT%H:%M:%S.%f'"}}
.CORS headers are now supported for every request (#160).
Wrong
subject
filter produces more user-friendly error (#162).Setting a keytab file is no longer required: if one is not explicitly set,
/etc/krb5.keytab
will be used (#55).Unused option
resultsdb_api_url
was removed from client.conf.Containers on Quay (https://quay.io/repository/factory2/waiverdb).
WaiverDB 0.9.0¶
Released 1 Mar 2018.
The usage of
JSONB
has been replaced with the olderJSON
column type, in order to maintain compatibility with PostgreSQL 9.2 on RHEL7 (#134).
WaiverDB 0.8.0¶
Released 16 Feb 2018.
Removed support to SQLite in favor of PostgreSQL.
Fixed database migration to use the correct column type for the
waiver.subject
column (#129).Added information on the README file on how to configure the db.
WaiverDB 0.7.0¶
Released 16 Feb 2018.
Fixed the database migration strategy for Openshift deployment (#121). The migration step is now run in a pre-deployment hook. Previously it ran in a post-start pod hook which did not work correctly in some situations.
WaiverDB 0.6.0¶
Released 13 Feb 2018.
Dummy authentication for CLI for developing and debugging reasons.
Added logo in the README page.
You can now waive the absence of a result. Now it is possible to submit waivers using a subject/testcase.
Backward compatibility for submitting a waiver using the result_id. This feature will be removed in the near future.
WaiverDB 0.5.0¶
Released 17 Jan 2018.
Database migrations have been introduced, and will be a part of future releases. Users upgrading to 0.5 will need to run these commands:
$ waiverdb db stamp 0a27a8ad723a $ waiverdb db upgrade
Error messages are now returned by the API in JSON format.
A new authentication method: ssl auth. See the docs for more on configuration.
The API now supports a proxyuser argument. A limited set of superusers, configured server-side, are able to submit waivers on behalf of other users.
WaiverDB 0.4.0¶
Released 08 Nov 2017.
A number of issues have been resolved in this release:
New WaiverDB CLI for creating waivers (#82).
New /about API endpoint to expose the current running version and the method used for authentication of the server.
Improved the process of building docs by using sphinxcontrib.issuetracker extension.
WaiverDB 0.3.0¶
Released 26 Sep 2017.
A number of issues have been resolved in this release:
Fixed some type errors in the API docs examples (#73).
Updated README to recommend installing package dependencies using dnf builddep (#74).
Fixed the health check API to return a proper error if the application is not able to serve requests (#75).
Other updates:
Supports a new HTTP API /api/v1.0/waivers/+by-result-ids.
Package dependencies are switched to python2-* packages in Fedora.
WaiverDB 0.2.0¶
Released 16 June 2017.
Supports containerized deployment in OpenShift.
DATABASE_PASSWORD
andFLASK_SECRET_KEY
can now be passed in as environment variables instead of being defined in the configuration file.Supports publishing messages over AMQP, in addition to Fedmsg. The
ZEROMQ_PUBLISH
configuration option has been renamed toMESSAGE_BUS_PUBLISH
.The
/etc/waiverdb/settings.py
configuration file is no longer installed by default. For new installations, you can copy the example configuration from/usr/share/doc/waiverdb/conf/settings.py.example
.Numerous improvements to the test and build process for WaiverDB.
WaiverDB 0.1.0¶
Initial release, 12 April 2017.