[AIRFLOW-7113] fix gantt render error

[cong-zhu]

---

Issue link: AIRFLOW-7113

Make sure to mark the boxes below before creating PR: [x]

• Description above provides context of the change
• Unit tests coverage for changes (not needed for documentation changes)
• Commits follow "How to write a good git commit message"
• Relevant documentation is updated including usage instructions.
• I will engage committers as explained in Contribution Workflow Example.

---

In case of fundamental code change, Airflow Improvement Proposal (AIP) is needed.
In case of a new dependency, check compliance with the ASF 3rd Party License Policy.
In case of backwards incompatible changes please leave a note in UPDATING.md.
Read the Pull Request Guidelines for more information.

[kaxil]

@cong-zhu Thanks for raising this PR. I know someone reported this error too. However, I couldn't reproduce the error. Do you have an example DAG or steps I can do to reproduce this error?

[KevinYang21]

@kaxil I think if we have tasks marked as failed by the scheduler upon receiving executor event then we'll end up having such failing tasks w/o a start date and crash the gantt chart.

@cong-zhu correct me if I'm wrong.

[KevinYang21]

The CI error looks irrelevant tho, maybe try rebase and rerun?

[cong-zhu]

@KevinYang21, yes it's correct.

@kaxil. Tasks fail at early stage, e.g. during parsing, won't contain the start date. When handling failures, scheduler will store those failed tasks in task_fail table. Which crashes the gantt chart. To reproduce the error, you need to have a row w/o start_date in task_fail table. Entries w/o start_date in task_instance table will be fine as the query filters them out. I reverted the change on it.

[kaxil]

@KevinYang21, yes it's correct.

@kaxil. Tasks fail at early stage, e.g. during parsing, won't contain the start date. When handling failures, scheduler will store those failed tasks in task_fail table. Which crashes the gantt chart. To reproduce the error, you need to have a row w/o start_date in task_fail table. Entries w/o start_date in task_instance table will be fine as the query filters them out. I reverted the change on it.

Shouldn't we change the Scheduler code to store the start_date and if there is any error store start_date = end_date during that time before we store in task_fail table?

[cong-zhu]

Shouldn't we change the Scheduler code to store the start_date and if there is any error store start_date = end_date during that time before we store in task_fail table?

I think it still makes sense to store the empty start_date as it indicates the task failed before running. We just need to appropriately represent this from the UI.