[GitHub] [airflow] potiuk commented on issue #17977: DagRun marked as success without executing task

[GitBox <gi...@apache.org>]

potiuk commented on issue #17977:
URL: https://github.com/apache/airflow/issues/17977#issuecomment-1336553800

   > Hi. We are facing the same issue on v2.3.4+composer. Is there any plan on fixing this any time soon? Is there any configuration we could review/change to prevent or mitigate this? We have many DAGs that depends on others to run, when they are marked as success then the depending DAGs are being executed but they did not find any data because of this problem.
   
   > Hi, we migrated from Google Cloud Platform Airflow 2.2.5+composer to Airflow 2.3.3+composer, and this change made our dags start failing with the description mentioned above. We have dynamic end dates, based on several threads (https://stackoverflow.com/questions/73652663/airflow-dag-run-marked-as-success-although-the-tasks-didnt-run) seem to be causing this bug. We have more than 100 dags, we would like to know if there is a planned fix for this issue, to decide what we will do next. We've tried v2.3.4 and the issue persists.
   
   Unless some of you submits additional information (logs, cicumstances, descriptions I don't think anything is going to happen - no-one will likely look at it, before we lack reproduction and information. The only way you can help us to speed up investigation and either fix, or helping you to diagnose bugs in your DAGs, is to provide more circumstancess - when it happens, examples of DAGs causing it (with specific information on how you generate start_date and the DAGs themselves). log files, observations, screenshots. 
   
   This issue was not commented on since Sep 2021 and it was orignally reported in 2.1.3. Also, the stack overflow issue mentions that this might be due to dynamic ``start_date`` (which is bad practice and you should fix it in your DAGs - look at our current examples and https://airflow.apache.org/docs/apache-airflow/stable/best-practices.html). So most likely than not, the issue for people reporting it were not following the best practices. It is likely the same in your case and you should follow this trail I think.
   
   Also, I am not sure where do you take from the information that there is a big and asking for "plans" to fix it. At this moment, we do not even know if this is a bug and certainly because of lack of that knowledge, there are no plans to fix it. 
   
   Also be cautious about your expectations. You are paid Composer user, and while I understand you might have certain expectations and SLAs there, this is an OSS project that comes "as is" and therare not "plans" and "expectations" towards bug fixes of individual users. It does not work like that. Things get fixed when someone fixes them. And sometime things reported here are not real issues so they don't get fixed. Or sometimes it's unclear - and if there are no clarity those things won't get fixed either.
   
   Some people here - users and maintainer - do their best to help users like you in their free time, but they do not provide any SLAs. Things are usually fixed if some member of the community (not necessarily maintainer) get enough information to diagnose them and we asses there are important to fix and decides to fix it. That's how the OSS works. 
   
   So the best you can do (both of you @enriquetuya  and @rdgfuentes ) is to try to provide enough informaiton that people who try to help you for free, can actually help to solve your problem.
   
   If you have some "expectations" and "SLA" - there are also companies and individuals who provide paid expertise for Airlfow - and maybe those will be better for you to approach if there are some expectations your business continuity depends on. That's much more "business" focused approach.
   
   Help us to help you, please if you want help here.
   


-- 
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.

To unsubscribe, e-mail: commits-unsubscribe@airflow.apache.org

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