Overview

Deep dive into Janus and its architecture

Janus is a set of custom Kubernetes resources that acts as a gatekeeper for filesystem permission events on specified paths. It provides a rich set of configurations to run alongside your existing Kubernetes deployments to ensure that outside processes can or cannot open or access these paths.

In deploying Janus into your cluster, a couple of components will be initialized. First, the controller listens directly for Kubernetes events and is responsible for syncing cluster state with the daemons running on each node to make sure specified pods that need to be guarded, are. Second, the daemon is run on each node to ensure that it can monitor running pods on the same node. It is charged with marking inodes via fanotify to allow you to set guards for certain filesystem events on the desired paths inside the pod.

We will provide some common patterns using sets of popular tools further in this documentation.

Architecture

graph RL A((procfs)) -->|fanotify| B[janusnotify] subgraph daemon C{janusd} --> B B .->|logfn| C end D[janus-controller] ==>|gRPC| C C .-> D subgraph controller E(K8s listers
K8s informers) --> D F(JanusGuard CRDs) --> D end

Custom Resource Definition

In order to describe the set of pods you want to create a guard for, what allow and deny paths you are interested in guarding, and for what events, you’ll soon find yourself creating a JanusGuard CustomResourceDefinition. That is, a Kubernetes custom type that will be configured in your cluster, that gives a specification to define a guard type with all the fields and flags the controller will understand, configure, and pass along to the daemon so it can guard appropriately in the fashion you describe.

Controller

janus-controller is the glue between the Kubernetes lifecycle and the daemons listening for events and sending permission responses on the filesystems. It responds to events such as pod listers add, update, and delete, making sure that the proper state of the daemon that is running on the same node as that pod has the guard either started or not on that daemon.

This controller operates in an idempotent fashion, in such cases that the controller pod running is terminated and a new one spins up, it immediately receives all the events upon startup that it would have received anyway over time; this way we can reconstruct the proper state immediately. When a daemon is not gracefully terminated, upon startup of a new daemon pod, we reconcile that pod’s current state (nothing) with what we expect it should be, and all addition calls should be immediately fired; same as the previous case but in reverse.

The other set of types the controller listens for is the JanusGuard CustomResourceDefinition described above. This is the definition that you will create to define what selector, paths, events, and optional flags to create the notion of file access auditing.

Daemon

The janusd daemon will be deployed on all cluster nodes via a DaemonSet. It starts a gRPC server to communicate back-and-forth with the controller, described above. When it receives a message from the controller to create a guard, it takes the desired configuration and spawns a janusnotify process with a custom set of rules. For example, if one wishes to guard the entire mountpoint that a directory or file resides on, this can be configured with an optional flag.

sequenceDiagram participant A as janusd participant B as janusnotify participant C as procfs A->>+B: Creates child processes B->>C: Creates fanotify marks Note over B,C: Polling for fanotify events... loop fanotify event C->>B: Write permission response opt exit signal B-->C: Stop polling and exit end end

Once the janusnotify process receives an fanotify message that the user configured to guard, it writes a response to allow or deny that event, based on how you configure your guard paths.

Upon receiving an additional create message, if the guard already exists, it will handle it in an idempotent way, and simply update with the current configuration. When receiving a delete message, if the guard exists, it will shut down the processes involved and remove this guard from the daemon’s current state.

There is an additional state check called periodically by the controller that reconciles what the daemon knows its state is with what the controller expects the current state to be. Any additional creation and deletion methods may be called due to any differences. This also allows for handling non-graceful terminations of either daemon or controller pods, thus the proper state can always be recreated in these cases.