Hi all! We started using the latest version of fleet and rancher in a greenfield environment. It's the first time using a GitOps approach to provision cluster. We do not do much as of now, but we already entangle our appbundles with dependencies making provisioning very brittle and error prone. For example: We want to deploy cert-manager into each cluster and we want to add a default certificate which provides a wildcard certificate for each cluster based on the cluster name. Creating this in a single bundle causes a race condition blocker, because fleet tries to apply the yaml for the certificate before the CRDs via cert-manager are introduced to the cluster. It seems that fleet doesn't "try again" but just stops with the provisioning of the bundle. We "resolved" that by creating a second bundle for the config (the certificate) with

dependsOn

pointing to the cert-manager install bundle. This is just one of the manual dependencies we used to resolve such conflicts. For some other installations, we ended up running 3-4 levels of

dependsOn

making things very much impractical. My gut feeling tells me, we are using fleet in a way it was not meant to be used or we understand the usage of fleet wrong. Do you have a recommendation on how to resolve / overcome these dependencies?