The JSON serializer for our API uses ``json.loads``, which permits *any*
valid JSON (including bare integers, boolean values, etc). Lots of our
code, however, assumes that inbound JSON content will be a dict.
see: #4756
* stable: (275 commits)
Install correct rpm-sign package in RPM builder
Updating changelog for 3.1 release
Switch job_type to check from sync when detecting delete_on_update
use Unicode apostrophes - not single quotes - for French i18n strings
pin appdirs==1.4.2
only cancel deps if we can cancel the inv update
fixing module_name check and adding support for the debug module
cancel jobs dependent on inv update
update tests
CSS tweaks to workflow results panels
like inventory updates, check if project update deps already processed
Revert "Merge pull request #5553 from chrismeyersfsu/fix-waiting_blocked"
Add awx/ui/client/languages to .gitignore
Delete awx/ui/client/languages/*.json
refactor based on review
Add missing permission check.
Make current_groups a set to easily avoid duplicates, update asgi-amqp requirement
avoid duplicated related search fields
Fix workflow audit items
fixing module name, json blob, and stdout-for-yum-module on host event
...
* release_3.1.0: (2069 commits)
Updating changelog for 3.1 release
Switch job_type to check from sync when detecting delete_on_update
use Unicode apostrophes - not single quotes - for French i18n strings
pin appdirs==1.4.2
only cancel deps if we can cancel the inv update
fixing module_name check and adding support for the debug module
cancel jobs dependent on inv update
update tests
CSS tweaks to workflow results panels
like inventory updates, check if project update deps already processed
Revert "Merge pull request #5553 from chrismeyersfsu/fix-waiting_blocked"
Add awx/ui/client/languages to .gitignore
Delete awx/ui/client/languages/*.json
refactor based on review
Add missing permission check.
Make current_groups a set to easily avoid duplicates, update asgi-amqp requirement
avoid duplicated related search fields
Fix workflow audit items
fixing module name, json blob, and stdout-for-yum-module on host event
just like we fail running tasks fail waiting tasks
...
* Associate the celery_id with the job at the earliest point possible.
This ensures that a waiting job has a celery id. Thus, we are free to
fail waiting jobs that don't have a celery id.
We need to not remove the source tree on project update run which
happens in the context of the job run to prevent the directory from
being deleted out from under the job runners. Since those will
inherently not block each other as the project update would
