[GitHub] [airflow] AyWa opened a new issue #9731: Task instance null, execution date skipped

[GitBox]

AyWa opened a new issue #9731:
URL: https://github.com/apache/airflow/issues/9731


   **Apache Airflow version**: `1.10.9`
   **DB**:  `Postgre11.6` (aws rds)
   
   **Environment**:
   
   - **Cloud provider or hardware configuration**: `AWS BEANSTALK - DOCKER` in 
a `Local` executor
   
   **What happened**:
   
   In the UI, everything looks to be correctly run:
   
   <img width="645" alt="Screen Shot 2020-07-09 at 1 04 12 PM" 
src="https://user-images.githubusercontent.com/13608477/86998728-f3347c80-c1eb-11ea-962b-d4c09c9cacac.png";>
   
   However, we figure out that one execution date was missing (in our 
application). So I tried to find the reason why, and in the tree view, they was 
no circle for this date.
   So I ran `SELECT * FROM dag_run where 
execution_date='2020-05-25T00:00:00+00:00'` and there was no row too.
   
   After that I check the pool task, and by filter by date (filter by status 
not equal to success was showing nothing), I found:
   
   <img width="1641" alt="Screen Shot 2020-07-09 at 1 04 01 PM" 
src="https://user-images.githubusercontent.com/13608477/86998972-7ce44a00-c1ec-11ea-9725-21433b181018.png";>
   
   So somehow, airflow UI is a bit misleading, because there was a task that 
failed to be run. But there was no easy way to figure out, without knowing the 
date or going to the database and running `SELECT * from task_instance where 
state is Null where  execution_date='2020-05-26T00:00:00+00:00';`
   
   ![Screen Shot 2020-07-09 at 1 24 37 
PM](https://user-images.githubusercontent.com/13608477/86999465-9e920100-c1ed-11ea-9dca-0f770acdb4cb.png)
   
   
   **What you expected to happen**:
   
   * I would expect to have an error or at least an indicator that a task 
instance are in null state or that a dag run was missing
   * I would expect to be able to filter `List Task Instance` by `NULL` or 
`None` status. However it was showing no result (if I was filtering by not 
equal to success it was empty too)
   * ideally I would expect to not have Null state, but If it is in the UI, I 
think it is acceptable
   
   **How to reproduce it**:
   
   Sadly I do not know how to reproduce, but I think by putting the data in the 
database we can at least "reproduce" the issue in the UI:
   
   - a dag run need to be empty `SELECT * FROM dag_run where 
execution_date='2020-05-25T00:00:00+00:00'`-> should give no row
   - task instance for this execution date should have an `Null` state like the 
screenshot shared. 
   
   Because we run in Elastic beanstalk, sometimes, container might be stop / 
recreated etc, so maybe it is related to that.


----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.

For queries about this service, please contact Infrastructure at:
us...@infra.apache.org