You are viewing a plain text version of this content. The canonical link for it is here.
Posted to commits@airflow.apache.org by "John Arnold (JIRA)" <ji...@apache.org> on 2018/04/18 19:43:00 UTC

[jira] [Created] (AIRFLOW-2340) SQLalchemy pessimistic connection handling not working

John Arnold created AIRFLOW-2340:
------------------------------------

             Summary: SQLalchemy pessimistic connection handling not working
                 Key: AIRFLOW-2340
                 URL: https://issues.apache.org/jira/browse/AIRFLOW-2340
             Project: Apache Airflow
          Issue Type: Bug
          Components: scheduler
    Affects Versions: 1.9.0
            Reporter: John Arnold
         Attachments: airflow_traceback.txt

Our scheduler keeps crashing, about once a day.  It seems to be triggered by a failure to connect to the postgresql database, but then it doesn't recover and crashes the scheduler over and over.

The scheduler runs in a container in our environment, so after several container restarts, docker gives up and the container stays down.

Airflow should be able to recover from a connection failure without blowing up the container altogether.  Perhaps some exponential backoff is needed?

 

See attached log from the scheduler.



--
This message was sent by Atlassian JIRA
(v7.6.3#76005)