When creating unified job, stash the list of pk values from the
instance groups returned from preferred_instance_groups so that the
task management system does not need to call out to this method
repeatedly.
.preferred_instance_groups_cache is the new field
- DependencyManager spawns dependencies if necessary
- WorkflowManager processes running workflows to see if a new job is
ready to spawn
- TaskManager starts tasks if unblocked and has execution capacity
- scm based inventory sources should launch project updates prior to
running inventory updates for that source.
- fixes scenario where a job is based on projectA, but the inventory
source is based on projectB. Running the job will likely trigger a
sync for projectA, but not projectB.
comments
- returns a special view to output the total number of children (and
grandchildren) events for all parents events for a job
value is the number of total children of that event
- intended to be consumed by the UI, as an efficient way to get the
number of children for a particular event
- see api/templates/api/job_job_events_children_summary.md for more info
Occasionally the create_partition will error with,
relation "main_projectupdateevent_20220323_19" already exists
This change wraps the db command into a try except block with its
own transaction
- ee organization can be changed to null (less restrictive)
- if organization is null, cannot be assigned to org (more restrictive)
- if org is assigned, it cannot be set to a different org
- Adds a Metrics() class that can track data such as number of
events the callback receiver inserted into database
- Exposes this metric data at the api/v2/metrics/ endpoint.
This data is prometheus-friendly
- Metric data is stored in memory, then periodically saved to Redis.
- Metric data is periodically broadcast to other nodes in the cluster,
so that each node has a copy of the most recent metric data collected.
Various points (e.g. created, running, processing events), are
structured into json format and output to /var/log/tower/job_lifecycle.log
As part of this work, the DependencyGraph is reworked to return
which job object is doing the blocking, rather than a boolean.
When targeting, ../workflow_job_templates/id#/workflow_nodes/ endpoint,
user could not set all_parents_must_converge to true.
3.7.1 backport for awx issue #7063
kubectl and synchronize are now part of community.kubernetes
and ansible.posix collections, respectively. This change installs
these collections to a local directory to be used in inventory and
isolated management playbooks.
awx issue #6930
With AWX_AUTO_DEPROVISION_INSTANCES on, instances
are registered with an ip address. However, new
instances might try to register before old instances
are deprivisioned. In this case old IPs can conflict with
the new ones. This will check for an ip conflict and unset
the IP of conflicting instance (set to None)
ansible/awx issue 6750
Fixed bug where an org admin was not able to add
an orphaned user to the org, in the case where the
orphan had an ancestor role that matched one of the
roles for of the org admin.
scenario to fix -- sue is member of cred1, where cred1 is
part of org1. org1 admin cannot add sue to org1, because
the cred1 role for sue has an ancestor to org1 role. The org1
admin cannot change or attach sue to org1.
tower issue #4198 and #4197
The commit is intended to speed up the cleanup_jobs command in awx. Old
methods takes 7+ hours to delete 1 million old jobs. New method takes
around 6 minutes.
Leverages a sub-classed Collector, called AWXCollector, that does not
load in objects before deleting them. Instead querysets, which are
lazily evaluated, are used in places where Collector normally keeps a
list of objects.
Finally, a couple of tests to ensure parity between old Collector and
AWXCollector. That is, any object that is updated/removed from the
database using Collector should be have identical operations using
AWXCollector.
tower issue 1103
Scenario - job is launched and spawns inventory update and project update.
If the inventory update fails, then it will fail the job and the project update.
It will fail the project update even if that update already ran and was successful.
This code change will not fail the project update if it has already ran successfully.
In cases where other jobs depend on that project update (but not the failed inventory
update), then we don't want those jobs to fail.
This adds a boolean "dependencies_processed" field to UnifiedJob
model. The default value is False. Once the task manager generates
dependencies for this task, it will not generate them again on
subsequent runs.
The changes also remove .process_dependencies(), as this method repeats
the same code as .process_pending_tasks(), and is not needed. Once
dependencies are generated, they are handled at .process_pending_tasks().
Adds a unit test that should catch regressions for this fix.
Versions selected to be pre-19 pip
due to unresolved issues with the build systems
Upgrade everything, party on
document new process
rotate license files
fix Swagger schema generation target
Remove --ignore-installed flag
Seth Foster