[Feature]: Provide insight in the current schedules in Metadata

[hanstrompert]

Contact Details

No response

What should we build?

For more information contact @wouter1975

Summary

Providing more details whether or not a schedule is made per workflow or task would be a good addition. The current implementation is that schedules are defined in the schedules folder in core or (surf/deployment) specific.

@scheduler(name="Validate NSO service models against subscriptions", time_unit="day", at="04:30")
def validate_nso_against_subscriptions() -> None:
    if not ProcessTable.query.filter(
        ProcessTable.workflow_name == "task_validate_nso_against_subscriptions",
        ProcessTable.last_status not in ["completed", "aborted"],
    ).count():
        start_process("task_validate_nso_against_subscriptions")

The information is not easily extractable from code, as the relevant task is specified by start_process("xxx"). The schedule frequency is defined by the decorator, by the keys, time_unit, at, period. This would translate to something like:

every period time_unit, like every 6 hours or based on specific time:

at at every time_unit, like at 0:10 every day

QUESTIONs:

• Can we get this information easily from the code and present this to the user in the FE, or should we consider changing this to a DB table with the bonus that adding/removing schedules is now also possible?
• Should we allow more flexibility, like 1 time only schedules?

Relevant pseudo code

[Mark90]

Proposal from brainstorm Hans and Mark: replace schedule library with apscheduler with SQLAlchemyJobStore as data store on the postgres backend. This would require changing the programmatic interface to defining a schedule for a task, and changing scheduler.py to run apscheduler.

The benefits would be:

• Having scheduler jobs in the datastore can be accessed from the API
• It becomes possible to schedule a "one-off" task from the UI. For instance if you want to run a task at 19:00 in the evening.
• Change/disable schedules defined in the core

[hanstrompert]

Scope for now:

• able to support current schedules
• store schedules in db
• make schedules available through api read-only