[D] I think making schedulers HA by using the support of NOWAIT in SQL is not a good solution [airflow]

["trikker (via GitHub)" <gi...@apache.org>]

GitHub user trikker created a discussion: I think making schedulers HA by using the support of NOWAIT in SQL is not a good solution

Hi, all:

I think making scheduler HA by using the support of SQL NOWAIT in metadata database is not a good solution because database need HA too. When DB HA switch happens, there may be the case that two schedulers are connected to different nodes of metadata database, and in this case, the SQL "select ... for update NOWAIT" may both succeed in two schedulers, and may finally make tasks scheduled twice ?

Is this right? Thanks for answering.

GitHub link: https://github.com/apache/airflow/discussions/44123

----
This is an automatically sent email for commits@airflow.apache.org.
To unsubscribe, please send an email to: commits-unsubscribe@airflow.apache.org