On large stacks (~400 resources) pulumi times out on some resources and exits, but doesn’t update the state with resources created in that time. Obviously on the next

pulumi up

it tries to created those resorces and fails as they already exist. It’s a big problem when this resource is an enabler for other resources (for example a k8s CRD controller) Is there a concurrency limit I can increase or any other way to solve this?

There's the

-p

flag, but based on the default value I don't think it will help you.

-p, --parallel int              Allow P resource operations to run in parallel at once (1 for no parallelism). Defaults to unbounded. (default 2147483647)