Large query results causing Redis to lose connection

[danielfaulknor]

Issue Summary

A very large number of results (960,000) causes Redis to give up during saving_results

Steps to Reproduce

1. Run a query that generates 960,000 rows
2. Watch it fail in the logs (https://gist.github.com/danielfaulknor/b6f6a99246cfcaa23e5020201d5c7fb8)

Why is this a bug in my view - This is constantly reproducible, and doesn't produce an error in the GUI, it just gets stuck on saving_results state. I have to run redis-cli flushall to clear the query.

The docker host has 12GB RAM and doesn't get past about 7.5GB in use.

Technical details:

• Redash Version: 5.0.2+b5486
• Browser/OS: Chrome on Windows
• How did you install Redash: docker-compose

[zachliu]

@danielfaulknor Have you tried to play with the CELERYD_MAX_TASKS_PER_CHILD(celery 3) or worker_max_memory_per_child(celery 4) parameter?