typhoon/docs/topics/security.md

58 lines
2.8 KiB
Markdown
Raw Normal View History

# Security
Typhoon aims to be minimal and secure. We're running it ourselves after all.
## Overview
**Kubernetes**
* etcd with peer-to-peer and client-auth TLS
Enable Kubelet TLS bootstrap and NodeRestriction * Enable bootstrap token authentication on kube-apiserver * Generate the bootstrap.kubernetes.io/token Secret that may be used as a bootstrap token * Generate a bootstrap kubeconfig (with a bootstrap token) to be securely distributed to nodes. Each Kubelet will use the bootstrap kubeconfig to authenticate to kube-apiserver as `system:bootstrappers` and send a node-unique CSR for kube-controller-manager to automatically approve to issue a Kubelet certificate and kubeconfig (expires in 72 hours) * Add ClusterRoleBinding for bootstrap token subjects (`system:bootstrappers`) to have the `system:node-bootstrapper` ClusterRole * Add ClusterRoleBinding for bootstrap token subjects (`system:bootstrappers`) to have the csr nodeclient ClusterRole * Add ClusterRoleBinding for bootstrap token subjects (`system:bootstrappers`) to have the csr selfnodeclient ClusterRole * Enable NodeRestriction admission controller to limit the scope of Node or Pod objects a Kubelet can modify to those of the node itself * Ability for a Kubelet to delete its Node object is retained as preemptible nodes or those in auto-scaling instance groups need to be able to remove themselves on shutdown. This need continues to have precedence over any risk of a node deleting itself maliciously Security notes: 1. Issued Kubelet certificates authenticate as user `system:node:NAME` and group `system:nodes` and are limited in their authorization to perform API operations by Node authorization and NodeRestriction admission. Previously, a Kubelet's authorization was broader. This is the primary security motivation. 2. The bootstrap kubeconfig credential has the same sensitivity as the previous generated TLS client-certificate kubeconfig. It must be distributed securely to nodes. Its compromise still allows an attacker to obtain a Kubelet kubeconfig 3. Bootstrapping Kubelet kubeconfig's with a limited lifetime offers a slight security improvement. * An attacker who obtains the kubeconfig can likely obtain the bootstrap kubeconfig as well, to obtain the ability to renew their access * A compromised bootstrap kubeconfig could plausibly be handled by replacing the bootstrap token Secret, distributing the token to new nodes, and expiration. Whereas a compromised TLS-client certificate kubeconfig can't be revoked (no CRL). However, replacing a bootstrap token can be impractical in real cluster environments, so the limited lifetime is mostly a theoretical benefit. * Cluster CSR objects are visible via kubectl which is nice 4. Bootstrapping node-unique Kubelet kubeconfigs means Kubelet clients have more identity information, which can improve the utility of audits and future features Rel: https://kubernetes.io/docs/reference/command-line-tools-reference/kubelet-tls-bootstrapping/ Rel: https://github.com/poseidon/terraform-render-bootstrap/pull/185
2020-04-26 01:50:51 +02:00
* Kubelets TLS bootstrap certificates (72 hours)
* Generated TLS certificate (365 days) for admin `kubeconfig`
* [NodeRestriction](https://kubernetes.io/docs/reference/access-authn-authz/node/) is enabled to limit Kubelet authorization
* [Role-Based Access Control](https://kubernetes.io/docs/admin/authorization/rbac/) is enabled. Apps must define RBAC policies for API access
* Workloads run on worker nodes only, unless they tolerate the master taint
2018-10-03 16:46:51 +02:00
* Kubernetes [Network Policy](https://kubernetes.io/docs/concepts/services-networking/network-policies/) and Calico [NetworkPolicy](https://docs.projectcalico.org/latest/reference/calicoctl/resources/networkpolicy) support [^1]
[^1]: Requires `networking = "calico"`. Calico is the default on all platforms (AWS, Azure, bare-metal, DigitalOcean, and Google Cloud).
**Hosts**
* Container Linux auto-updates are enabled
* Hosts limit logins to SSH key-based auth (user "core")
**Platform**
* Cloud firewalls limit access to ssh, kube-apiserver, and ingress
* No cluster credentials are stored in Matchbox (used for bare-metal)
* No cluster credentials are stored in Digital Ocean metadata
* Cluster credentials are stored in AWS metadata (for ASGs)
2018-08-27 08:39:41 +02:00
* Cluster credentials are stored in Azure metadata (for scale sets)
* Cluster credentials are stored in Google Cloud metadata (for managed instance groups)
* No account credentials are available to Digital Ocean droplets
2018-08-27 08:39:41 +02:00
* No account credentials are available to AWS EC2 instances (no IAM permissions)
* No account credentials are available to Azure instances (no IAM permissions)
* No account credentials are available to Google Cloud instances (no IAM permissions)
## Precautions
Typhoon limits exposure to many security threats, but it is not a silver bullet. As usual,
* Do not run untrusted images or accept manifests from strangers
* Do not give untrusted users a shell behind your firewall
* Define network policies for your namespaces
## Container Images
Typhoon uses upstream container images (where possible) and upstream binaries.
!!! note
Kubernetes releases `kubelet` as a binary for distros to package, either as a DEB/RPM on traditional distros or as a container image for container-optimized operating systems.
Typhoon [packages](https://github.com/poseidon/kubelet) the upstream Kubelet and its dependencies as a [container image](https://quay.io/repository/poseidon/kubelet) for use in Typhoon. The upstream Kubelet binary is checksummed and packaged directly. Quay automated builds provide verifiability and confidence in image contents.
## Disclosures
If you find security issues, please email dghubble at gmail. If the issue lies in upstream Kubernetes, please inform upstream Kubernetes as well.