ebdbec99eb
Currently, it is running only as part of the watcher-api when running as an standalone eventlet service, and not executed as a wsgi service. In a decision-engine failure scenario, the continuous audits running on it are not reassigned and are not longer running. This patch is moving the service to be part of the decision-engine itself. Note that when there are no running decision-engines there is no point in running the monitor service. I am also including a leader election mechanism so that we can run multiple copies of the monitor and only one of the alive ones takes care of monitoring the service and acting upon failures. Assisted-By: claude-code Closes-Bug: #2126767 Change-Id: If13e8ebbfd632b98742b965299b8633ff894b212 Signed-off-by: Alfredo Moralejo <amoralej@redhat.com>
18 lines
769 B
YAML
18 lines
769 B
YAML
fixes:
|
|
- |
|
|
When running watcher-api as a wsgi server, the decision-engine monitor
|
|
service was not executed so, in a decision-engine failure scenario, the
|
|
continuous audits running on it are not reassigned and are not longer
|
|
running.
|
|
This patch moves the monitor service to the decision-engine so, it will
|
|
be executed as soon as any decision-engine is running.
|
|
deprecations:
|
|
- |
|
|
The APISchedulingService has been removed from the Watcher API service.
|
|
It is replaced functionally by the ServiceMonitoringService included
|
|
in the watcher-decision-engine.
|
|
features:
|
|
- |
|
|
A leader election mechanism has been added to the ServiceMonitoringService
|
|
so that multiple copies of the decision-engine can be executed simultaneously.
|