[question] What are the risks of the "experimental" generate_node_id feature?
jwietelmann opened this issue · 4 comments
Seems like generate_node_id
is a couple years old now. Is it considered experimental?
We're looking at using Kubernetes autoscaling, so to handle the issue of what happens when we scale down, it's either use generate_node_id
or migrate off of verk
. I'd like to figure out how safe the feature is at this point.
What should we be watching out for, if anything?
Great question. I think we can safely update this to not be considered experimental anymore. It has been used in production for a few years and it has worked as intended 👌 . It could probably default to true
tbh
Does this mean if I have generate_node_id
to true
, I can safely spin up two dynos on Heroku for my app?
Thanks in advance and thanks for creating verk
Yes! It's the way to go!
That's great! Thanks again!