Running a Node in ECS Results in ~2min delays during initialization

[virgofx]

Attempting to run apache/superset:latest in an Amazon ECS cluster. When starting a node we are observing a 2 minute delay when starting any Flask task. Just after the INFO:root:Configured event logger of type <class 'superset.utils.log.DBEventLogger'> the app will hang/freeze for around 2 minutes before continuing normally.

Is there any more debug I could turn on to try to figure out the issue with the painfully slow startup?

• RDS configured with Postgres 14
• Elasticache configured with Redis
• Running a single celery node with basic config

Timestamp	Log
2023-05-31T15:20:55.418	'FLASK_ENV' is deprecated and will not be used in Flask 2.3. Use 'FLASK_DEBUG' instead.
2023-05-31T15:20:55.428	logging was configured successfully
2023-05-31T15:20:55.428	INFO:superset.utils.logging_configurator:logging was configured successfully
2023-05-31T15:20:55.431	INFO:root:Configured event logger of type <class 'superset.utils.log.DBEventLogger'>
observe 2 min delay
2023-05-31T15:23:07.012	/usr/local/lib/python3.9/site-packages/flask_appbuilder/models/sqla/interface.py:67: SAWarning: relationship 'Tag.objects' will copy column tag.id to colu

When testing locally via Docker compose and a single node, cache node, and database node, the time between these calls is milliseconds.

[virgofx]

Figured out the issue - Hopefully this might help others who experience same issue.

Right after it initializes the logger, the app calls out to the database. In our case we had a dual stack network where DNS would return both IPv6 and IPv4 but would first try IPv6. IPv6 had a black hole which resulted in database connectivity hanging on a socket that would't resolve for around 2 minutes.