[Filters] Add filtering support to Search - Dags

[cmarteepants]

View

Search - Dags

Filters to Add

• Filter DAGs with Import Errors
  ↪ has_import_errors = true

• Filter DAGs by Owner
  ↪ owners = ['alice', 'bob', 'airflow']

• Filter Stale DAGs
  ↪ is_stale = true

• Filter DAGs by Timetable
  ↪ (e.g. CronTriggerTimetable, AssetOrTimeSchedule)

• Filter DAGs by Bundle Name or Version
  ↪ bundle_name = 'analytics', bundle_version = 'v1.2.3'

• Filter DAGs with Asset-Based Schedules
  ↪ Requires awareness of upstream asset dependencies

• Filter Unscheduled DAGs
  ↪ schedule_interval = None or equivalent

API Support

⚠️ Partial

• /dags returns many required fields (has_import_errors, is_stale, bundle_name, owners)
• But does not currently support these as filter query parameters
• Would require extending the DAG API to support additional filters

UI Impact

• Filters would appear at the top of the Search – DAGs screen
• Should be consistent with filters for status, tags, and active/paused state already present
• Multi-select or autocomplete where applicable (e.g., owners)

[AryanK1511]

I would love to work on this issue.

[AryanK1511]

Thank you for assigning this to me. I will start working on it now.

[AryanK1511]

Hi! I had a question and would appreciate your input on this.

You mentioned wanting a UI filter to show only DAGs with import errors (has_import_errors = true). That makes sense, and I actually opened this issue to implement the change on the API side as well.

However, after digging into it, I realized there’s a complication: even though the Dags table has a has_import_errors field (which is false when there are no errors), if a DAG does have an import error, it never gets an entry in the Dags table at all. I tested this by creating a DAG with an import error:

from airflow import DAG
from datetime import datetime

# This will break the import!
this_will_throw_an_error

default_args = {
    'start_date': datetime(2023, 1, 1),
}

dag = DAG(
    'broken_dag',
    default_args=default_args,
    schedule_interval=None,
)

As expected, the database only has my older (valid) DAG, but not the new one with the import error.

The broken DAG does show up in the Airflow UI (in the "Import Errors" section), but there’s no entry for it in the Dags table in the DB.

This seems to be because there’s a separate import_error table handling these cases. I’ve also confirmed this with @rawwar .

Given this setup, it seems like adding an API/UI filter on has_import_errors wouldn’t actually be useful, since the broken DAGs aren’t in the Dags table to be filtered in the first place, they’re already shown separately in the UI.

Could you please advise whether we should still go ahead with this filter, or if it makes more sense to close the issue I opened yesterday as "not planned"? I can start looking into other filters in the meantime. Would appreciate a quick response, thanks!

[pierrejeambrun]

Yes maybe just disregard that field for now and focus on other filters.

It’s weird I wasn’t able to identify where that Dag attribute was set. I’ll need to do more digging.

[cmarteepants]

Thanks @AryanK1511 - sorry for the delay. Agree that that's weird. If you come across more that don't work or can't get to any, just comment on the issue and move on. All and any progress is helpful, happy to create other issues if we need to for tracking purposes.

Thanks again for taking this on!

[AryanK1511]

Thank you so much for your help guys. I will start working on the other stuff now.