Red Hat turns air-gapped Dev Spaces rollout traps into a platform checklist

Red Hat has published a production-focused guide for scaling OpenShift Dev Spaces in restricted networks. The article is less an installation walkthrough than a catalog of failure modes that tend to appear after a proof of concept: injected sidecars, rootless-container permissions, configuration drift, weak capacity signals and hidden calls to public services.

Runtime integrations can break workspaces first

The guide identifies four common sources of friction: third-party monitoring agents, rootless container runtimes, custom install namespaces and admission webhooks. One concrete example is Dynatrace OneAgent injection. Its init container can alter file permissions and volume mounts in ways that prevent Dev Spaces workspace pods from starting. Red Hat’s prescribed mitigation is to exclude the Dev Spaces control and workspace namespaces with dynatrace.com/inject: "false" rather than debug the failure inside each workspace.

Container builds need a different exception. To run rootless Podman inside a workspace, the CheCluster configuration enables container-run capabilities and assigns the container-build and container-run security context constraints. The sample also grants SETUID and SETGID plus privilege escalation for user-namespace mapping, while keeping privileged containers disabled.

The article documents a version-adjacent trap surfaced by Dev Spaces 3.26: an older script assigned a subordinate UID/GID range of 2,147,483,646 even though a Linux user namespace is limited to 65,536 IDs. Stricter enforcement exposed the latent error. The remedy depends on the Universal Developer Image version and uses a compliant range rather than treating the failure as a generic Podman regression.

Disconnected means mirroring more than containers

Red Hat reduces air-gapped readiness to five dependency classes that must be resolved before onboarding developers. Platform teams must mirror the Operator catalog and Dev Spaces images, route public image sources through an ImageDigestMirrorSet, run an internal Open VSX extension marketplace, host IDE binaries internally and disable the default public devfile registry. Internal certificate authority trust must also reach workspace Git operations.

The readiness table turns those requirements into tests: a ready CatalogSource, a pod pulling the Universal Developer Image from the mirror, an internal Open VSX URL in the CheCluster, configured internal IDE download URLs, a dashboard free of internet-access errors and a successful git clone to an internally signed Git host.

Treat the configuration as a product

The guide recommends managing manifests through CI/CD rather than ad hoc oc patch commands or console edits. It couples that declarative path with Prometheus ServiceMonitor resources and proactive capacity alerts, so platform teams can see whether the shared development service is approaching resource limits before workspace startups fail.

The useful lesson is that an air gap is not a registry setting. Dev Spaces depends on catalogs, base images, extensions, editor downloads, devfiles, certificates and cluster policies; a viable internal developer platform has to make each dependency explicit and testable.