Inactive jobs cleanup
Inactive jobs clean up may be useful in the case of many jobs (which required a lot of Kubernetes cluster resources) supposed to be created and will not be in demand during some time. To free Kubernetes cluster resources it is possible to enable automatic inactive jobs clean up.
- Requirements:
LUNA_SERVICE_METRICS setting to be enabled
LUNA_LAMBDA_INACTIVE_JOB_CLEAR_INTERVAL setting to be enabled and properly configured
When active parameter in LUNA_LAMBDA_INACTIVE_JOB_CLEAR_INTERVAL setting is set to true, it will check every check_interval seconds and automatically disable jobs which last_usage is older than specified interval (taking interval_type into account). If job is newly created or updated, it will not remove such job even in case it’s last_usage is NULL until job last_update_time is not older than specified interval.
Note
Eternal jobs, jobs which creation are in progress and jobs which creation were failed will not be taken into account during the above procedure.