Each sidecar / init container requires some standard pre-configuration, and usually requires changes to:
Conjur policy
Kubernetes role bindings
Kubernetes config maps for Conjur SSL certificate
Kubernetes config maps for additional configuration files (eg secretless.yml)
Updated deployment manifests that includes the configuration for the sidecar/init container to communicate with Conjur
There are improvements that can be made to make the process of deploying applications in Kubernetes / OpenShift at scale that are set up to use the appropriate sidecar and communicate securely and with least privilege with Conjur. To that end, we developed the Conjur sidecar injector - a mutating admission webhook controller that can dynamically inject sidecars at application deploy time.
In this effort, we will explore modern options for best-practice management of sidecar automation, with the goal of coming up with a proposed plan for a path forward for the management of our Kubernetes integrations, with a desired end state and incremental stages to reach that end state. In developing this plan, we will likely spike out some options and develop some lightweight proofs-of-concept of some of the pieces we propose.
At current we have multiple sidecars that can be deployed with applications running in Kubernetes / OpenShift to facilitate connections to Conjur:
Each sidecar / init container requires some standard pre-configuration, and usually requires changes to:
There are improvements that can be made to make the process of deploying applications in Kubernetes / OpenShift at scale that are set up to use the appropriate sidecar and communicate securely and with least privilege with Conjur. To that end, we developed the Conjur sidecar injector - a mutating admission webhook controller that can dynamically inject sidecars at application deploy time.
In this effort, we will explore modern options for best-practice management of sidecar automation, with the goal of coming up with a proposed plan for a path forward for the management of our Kubernetes integrations, with a desired end state and incremental stages to reach that end state. In developing this plan, we will likely spike out some options and develop some lightweight proofs-of-concept of some of the pieces we propose.