6 ms·
Dokku maintainer here. You may wish to look into a message processing system in your language of choice and run that as a daemon in Dokku. We have plugins for
by josegonzalez 2y ago
Dokku maintainer here.
You may wish to look into a message processing system in your language of choice and run that as a daemon in Dokku. We have plugins for various datastores, and commonly I see folks just connect their worker processes to that. It's much lighter weight than spawning new containers for each workload.
- hedgehog 2y agoIn my case spawn overhead is negligible relative to the jobs themselves, my existing experience with AWS Batch has been pretty good. I took another look at the Dokku documentation and extending `run:detached` would work well enough, though maybe it's time for me to revisit Airflow or something in that direction.