**Autocert** is a kubernetes add-on that automatically injects TLS/HTTPS certificates into your containers.
<!--- TODO: 👋 Welcome. We ❤️ feedback. Submit an issue. Fork and send a PR. Give us a ⭐ if you like what we're doing. --->
<!--- 👋 Welcome. We ❤️ feedback. Submit an issue. Fork and send a PR. Give us a ⭐ if you like what we're doing. --->
**Autocert** is a kubernetes add-on that automatically injects TLS/HTTPS certificates into your containers.
To get a certificate **simply annotate your pods** with a name. An X.509 (TLS/HTTPS) certificate is automatically created and mounted at `/var/run/autocert.step.sm/` along with a corresponding private key and root certificate (everything you need for [mTLS](#motivation)).
@ -23,51 +23,37 @@ TODO: Twitter, Slack, Issues (tagged with #autocert / special template)...
* [Getting a certificate locally](#local-certificate)
* [How it works](#how-it-works)
* [Uninstalling](#uninstalling)
* [Questions](#questions)
## Motivation
`Autocert` exists to **make it easy to use mTLS** (mutual TLS) to **improve security** within a cluster and to **secure communication into, out of, and between kubernetes clusters**. The goal is to make right and proper public key infrastructure (PKI) more accessible to people running kubernetes.
## Features
TLS (and HTTPS, which is HTTP over TLS) provides _authenticated encryption_: an _identity dialtone_ and _end-to-end encryption_ for your workloads. It's like a secure line with caller ID. This has all sorts of benefits: better security, compliance, and easier auditability for starters. It makes workloads identity-aware, improving observability and enabling granular access control. Perhaps most compelling, [mutual TLS](#) (mTLS) lets you securely communicate with workloads running anywhere, not just inside kubernetes.
Autocert uses [`step certificates`](https://github.com/smallstep/certificates) to generate keys and issue certificates from your own **internal certificate authority**. This process is secure and automatic, all you have to do is [install autocert](#install) and [annotate your pods](#annotate-pods). Features include:
TODO: Diagram
* A complete **public key infrastructure** (PKI) for your kubernetes clusters
* A fully featured internal **certificate authority** (CA) that you control so you can **use mTLS to control access to services**
* Ability to run subnordinate to or federated with an existing PKI
* CA and PKI artifacts are installed in their own namespace (`step`) for easy access control
* Modern certificate best practices
* Automated certificate management (auto enrollment and renewal)
* Short-lived certificates
* Private keys are never transmitted across the network (and aren't stored in `etcd`)
* RFC5280 and CA/Browser Forum compliant certificates that work with browsers and other standard TLS implementations
* Builds on [`step certificates`](https://github.com/smallstep/certificates) so you can also issue certificates to servers, people, and code running in a different cluster and outside of kubernetes
If you know how to operate and scale DNS and proxy infrastructure then you already know how to scale and operate secure service-to-service communication using mTLS (mutual TLS). There's just one problem: **you need certificates issued by your own certificate authority (CA)**. Building and operating a CA, issuing certificates, and making sure they're renewed before they expire is tricky. `Autocert` does all of this for you.
## Motivation
Because `autocert` is built on [`step certificates`](#) you can easily extend access to developers, endpoints, and workloads running outside your cluster.
TLS (e.g., HTTPS) is the most widely deployed cryptographic protocol in the world. Mutual TLS (mTLS) provides end-to-end security for service-to-service communication and can **replace complex VPNs** to secure communication into, out of, and between kubernetes clusters. But **to use mTLS to secure internal services you need certificates issued by your own certificate authority (CA)**.
## Features
Building and operating a CA, issuing certificates, and making sure they're renewed before they expire is tricky. Autocert does all of this for you.
First and foremost, `autocert` is easy. You can **get started in minutes**.
## Getting Started
`Autocert` uses [`step certificates`](https://github.com/smallstep/certificates) to generate keys and issue certificates. This process is secure and automatic, all you have to do is [install autocert](#install) and [annotate your pods](#annotate-pods).
Features include:
These instructions will get `autocert` installed quickly on an existing kubernetes cluster.
* A fully featured private **certificate authority** (CA) for workloads running on kubernetes and elsewhere
* [RFC5280](#) and [CA/Browser Forum](#) compliant certificates that work **for TLS**
* Namespaced installation into the `step` namespace so it's **easy to lock down** your CA
* Short-lived certificates with **fully automated** enrollment and renewal
* Private keys are never transmitted across the network and aren't stored in `etcd`
## Getting Started
### Prerequisites
You'll need `kubectl` and a kubernetes cluster running version `1.9` or later with [webhook admission controllers](https://kubernetes.io/docs/reference/access-authn-authz/extensible-admission-controllers/#admission-webhooks) enabled:
All you need to get started is [`kubectl`](https://kubernetes.io/docs/tasks/tools/install-kubectl/#install-kubectl) and a cluster running kubernetes `1.9` or later with [admission webhooks](https://kubernetes.io/docs/reference/access-authn-authz/extensible-admission-controllers/#admission-webhooks) enabled:
For [manual installation](INSTALL.md) you'll also need to [`install step`](https://github.com/smallstep/cli#installing) version `0.8.3` or later.
```bash
$ step version
Smallstep CLI/0.8.3 (darwin/amd64)
Release Date: 2019-01-16 01:46 UTC
```
### Install
To install `step certificates` and `autocert` in one step run:
To install `autocert` run:
```bash
$ kubectl run autocert-init -it --rm --image smallstep/autocert-init --restart Never
kubectl run autocert-init -it --rm --image smallstep/autocert-init --restart Never
```
The init script will end by printing:
💥 installation complete.
* The admin provisioner password (also used to encrypt the CA's key material)
* The `autocert` provisioner password (used by the mutating webhook)
* Your CA's root certificate fingerprint (used to bootstrap secure communication)
> You might want to [check out what this command does](init/autocert.sh) before running it. You can also [install `autocert` manually](INSTALL.md) if that's your style.
Feel free to store these some place safe. The passwords are also stored as secrets in the `step` namespace.
## Usage
> 🤔 **Tip:** If you lose your root certificate fingerprint you can calculate it again by running:
To tell `autocert` to issue certificates to a pod's containers you need to [specify a name](RUNBOOK.md#naming-considerations) using the `autocert.step.sm/name` annotation. This name will appear in the issued certificate (as the X.509 common name and SAN).
For `autocert` to inject a certificate pods must use the `autocert.step.sm/name` annotation to specify their name. The value of this annotation will appear as the name in the issued certificate (the X.509 common name and SAN).
To test your installation apply this annotated deployment, which starts a [simple server](examples/hello-mtls/go/server.go) that uses mTLS:
Once the pod has started we can check that our certificate, private key, and root certificate have been properly mounted in our container at `/var/run/autocert.step.sm`.
Our new container should have a certificate, private key, and root certificate mounted at `/var/run/autocert.step.sm`:
```bash
$ export SLEEP_POD=$(kubectl get pods -l app=sleep \
-o jsonpath={$.items[0].metadata.name})
$ kubectl exec -it $SLEEP_POD -c sleep -- ls /var/run/autocert.step.sm
$ export HELLO_MTLS=$(kubectl get pods -l app=hello-mtls -o jsonpath={$.items[0].metadata.name})
$ kubectl exec -it $HELLO_MTLS -c hello-mtls -- ls /var/run/autocert.step.sm
> 🤔 **Tip:** The `autocert-renewer` sidecar also installs the [`step` CLI tool](https://github.com/smallstep/cli), which we can use to inspect the issued certificate.
With `autocert` issuing and rotating certificates we can start using mTLS between services. The [`examples/hello-mtls`](examples/hello-mtls) directory demonstrates the right way to do mTLS in several languages (contributions welcome :). Let's deploy one.
### Mutual TLS
Build and deploy the `hello-mtls` server for golang:
We're done. Our container has a certificate, issued by our CA, and `autocert` will take care of renewal automatically.
It's easy to deploy certificates automatically with `autocert`, but it's up to you to use them correctly. To get you started, [`hello-mtls`](examples/hello-mtls) demonstrates the right way to use mTLS with various tools and languages (contributions welcome :). If you're a bit fuzzy on how mTLS works, [the `hello-mtls` README](examples/hello-mtls) is a great place to start.
Check that the client is connecting and working as expected:
To finish out this tutorial let's keep things simple and try `curl`ing the server we just deployed from inside and outside the cluster.
```bash
$ export HELLO_MTLS=$(kubectl get pods -l app=hello-mtls-client \
Once deployed the client logs should indicate that it's successfully connecting to our server using mTLS, which is [echoing the client's name](examples/hello-mtls/go/server.go#L71-L72) in response.
> curl: (60) SSL certificate problem: unable to get local issuer certificate
> More details here: https://curl.haxx.se/docs/sslcerts.html
>
> curl failed to verify the legitimacy of the server and therefore could not
> establish a secure connection to it. To learn more about this situation and
> how to fix it, please visit the web page mentioned above.
> ```
>
> You'll get similar errors from other tools, libraries, and applications if they're not properly configured to use the `autocert` certificates and keys. Minimally, for (non-mutual) TLS:
>
> * Clients must be configured to trust the `autocert` root certificate (`/var/run/autocert.step.sm/root.crt`) to authenticate a server
> * Servers must be configured to use the key and certificate issued by `autocert` (`/var/run/autocert.step.sm/site.crt` and `/var/run/autocert.step.sm/site.key`) to authenticate *to* a client
>
> If you're doing mTLS the inverse is also true: the server must trust the root certificate to authenticate client, and the client must be configured to use the `autocert`-issued key and certificate. In other words, for mTLS both the client and server should be configured to use `autocert`'s `root.crt`, `site.crt`, and `site.key`. With `curl` this is done using the `--cacert`, `--cert`, and `--key` flags, respectively.
### Exposing services using mTLS
With properly configured mTLS, services can be safely exposed directly to the public internet: **only clients that have a certificate issued by the internal certificate authority will be allowed to connect**. To demonstrate let's expose our `hello-mtls` service.
If you need a refresher, here's a rough approximation of how an mTLS handshake works:
* It's the signing of random numbers that proves we're talking to the right remote. It's the digital equivalent of asking someone to send you a photo of them with today's newspaper.
* The client and server need to have prior knowledge of the root certificate(s) used for signing other certificates.
* The client and server need to be configured to use the correct certificate and private key (the certificate must have been issued by a CA with a trusted root certificate)
#### Exposing `hello-mtls`
Because `hello-mtls` does proper mTLS itself we can expose it simply using a [service with type LoadBalancer](https://kubernetes.io/docs/concepts/services-networking/service/#loadbalancer).
For kicks, let's `exec` into this pod and try `curl`ing ourselves:
To connect to `hello-mtls` from outside kubernetes we need a certificate issued by our internal CA. Since `autocert` is built on `step certificates` we can securely issue certificates to users, devices, and workloads running in other environments.
Connecting from outside the cluster is a bit more complicated. We need to handle DNS and obtain a certificate ourselves (tasks which were handled automatically inside the cluster by kubernetes and `autocert`, respectively).
> 🤯 **Note:** To follow along you'll need `step` [installed locally](https://github.com/smallstep/cli#installing).
That said, since we're using mTLS our server can be safely exposed directly to the public internet using a [LoadBalancer service type](https://kubernetes.io/docs/concepts/services-networking/service/#loadbalancer): **only clients that have a certificate issued by our certificate authority will be allowed to connect**.
First, port-forward from localhost to the `step-ca` pod:
```
$ export CA_POD=$(kubectl -n step get pods -l app=ca \
To connect we need a certificate, which we'll need to obtain from the CA. There are a [couple](RUNBOOK.md#federation) [ways](RUNBOOK.md#exposing-the-ca) to do this, but for simplicity we'll just forward a port.
To follow along you'll need to [`install step`](https://github.com/smallstep/cli#installing).
Now we can use `step` to securely grab the CA's root certificate and obtain a certificate. You'll be prompted to select a provisioner and enter the correct password to continue:
In another window we can use `step` to grab the root certificate, generate a key pair, and get a certificate to use with `curl`. You'll need the admin password and CA fingerprint output during installation (see [here](RUNBOOK.md#recovering-fingerprint) and [here](#RUNBOOK.md#recovering-admin-password) if you already lost them :).
> 🤔 **Tip:** If you want someone (or something) to have a certificate with a particular name, but don't want to give them the ability to provision arbitrary certificates, you can generate a bootstrap token for them:
>
> ```bash
> $ step ca token snarf.local.dev \
> --ca-url https://127.0.0.1:4443 \
> --root root.crt
> eyJhbG...
> ```
>
> They can use the token to obtain a certificate (once):
>
> ```bash
> $ step ca certificate snarf.local.dev snarf.crt snarf.key --token "eyJhbG..."
> ```
>
> Actually, this is exactly what the `autocert` mutating webhook is doing for your pods! Read [how it works](#how-it-works) for more info.
#### Connecting to `hello-mtls`
We're ready to securely connect to `hello-mtls`.
Now we can simply `curl` the service:
```
$ export HELLO_MTLS_IP=$(kubectl get svc hello-mtls-lb -ojsonpath={$.status.loadBalancer.ingress...?})
$ export HELLO_MTLS_IP="127.0.0.1"
$ export HELLO_MTLS_IP=$(kubectl get svc hello-mtls-lb -ojsonpath={$.status.loadBalancer.ingress[0].ip})
> 🤯 **Note:** HTTPS clients check that the name in the server's cerificate match the `authority` portion of the URL (e.g., `https://smallstep.com/` must present a certificate with the name `smallstep.com`). (See [RFC2818](https://tools.ietf.org/html/rfc2818#section-3).)
>
> Our `hello-mtls` service's certificate binds the name `hello-mtls.default.svc.cluster.local` so we *must* connect to it using that name. If we use a different authority we'll get an error:
>
> ```
> $ curl --cacert root.crt \
> --cert snarf.crt \
> --key snarf.key \
> https://127.0.0.1
> curl: (51) SSL: no alternative certificate subject name matches target host name '127.0.0.1'
> ```
>
> In a real production environment you'd address this by either:
>
> * using a properly registered domain name and configuring DNS either globally (e.g., using [ExternalDNS](https://github.com/kubernetes-incubator/external-dns/)), or
> * using internal names and configuring DNS locally in each environment (e.g., using an [ExternalName service](https://kubernetes.io/docs/concepts/services-networking/service/#externalname))
>
> In any case, `hello-mtls.default.svc.cluster.local` must resolve to the right IP.
>
> You could use `/etc/hosts`. Since we're testing with `curl` it's even easier to use the `--resolve` flag to override resolution for a single request.
> If you're using minikube or docker for mac the load balancer's "IP" might be `localhost`, which won't work. In that case, simply `export HELLO_MTLS_IP=127.0.0.1` and try again.
> Note that we're using the `--resolve` flag to tell `curl` to resolve the name in our workload's certificate to its public IP address. In a real production infrastructure you could configure DNS manually, or you could propagate DNS to workloads outside kubernetes using something like [ExternalDNS](#).
✅ mTLS outside cluster.
### Cleanup & uninstall
To clean up after running through the tutorial remove the `hello-mtls` and `hello-mtls-client` deployments and services:
```
kubectl delete deployment hello-mtls
kubectl delete deployment hello-mtls-client
kubectl delete service hello-mtls
kubectl delete service hello-mtls-lb
```
The runbook contains instructions for [uninstalling `autocert` complete](RUNBOOK.md#uninstalling).
<!--- TODO: CTA or Further Reading... Move "How it works" maybe? Or put this below that? --->
## How it works
### Architecture
`Autocert` consists of a [webhook admission controllers](https://kubernetes.io/docs/reference/access-authn-authz/extensible-admission-controllers/#admission-webhooks) that injects one init container and one sidecar container to handle obtaining a certificate for the first time and renewing a certificate, respectively.
The `autocert` admission webhook will intercept this pod creation request and inject an [init container](bootstrapper/) and [sidecar](renewer/) to manage certificate issuance and renewal, respectively.
`Autocert` is an [admission webhook](https://kubernetes.io/docs/reference/access-authn-authz/extensible-admission-controllers/#admission-webhooks) that intercepts and patches pod creation requests with [some YAML](install/02-autocert.yaml#L26-L44) to inject an [init container](bootstrapper/) and [sidecar](renewer/) that handle obtaining and renewing certificates, respectively.
It integrates with [`step certificates`](https://github.com/smallstep/certificates) and uses the single-use token bootstrap protocol from that project to mutually authenticate a new pod with your certificate authority.
It integrates with [`step certificates`](https://github.com/smallstep/certificates) and uses the [one-time token bootstrap protocol](https://smallstep.com/blog/...) from that project to mutually authenticate a new pod with your certificate authority, and obtain a certificate.
Tokens are [generated by the admission webhook](controller/provisioner.go#L46-L72) and [transmitted to the injected init container via a kubernetes secret](controller/main.go#L91-L125). The init container [uses the one-time token](bootstrapper/bootstrapper.sh) to obtain a certificate. A sidecar is also installed to [renew certificates](renewer/Dockerfile#L8) before they expire. Renewal simply uses mTLS with the CA.
* Link to ExternalDNS example
* Link to multiple cluster with Service type ExternalDNS so they can communicate
## Questions
### Uninstall
#### How is this different than [`cert-manager`](https://github.com/jetstack/cert-manager)
* Delete the `sleep` deployment (if you created it)
* Remove labels (show how to find labelled namespaces)
* Remove annotations (show how to find any annotated pods)
* Remove secrets (show how to find labelled secrets)
* Delete `step` namespace
`Cert-manager` is a great project. However, it's designed primarily for managing certificates issued by [Let's Encrypt's](https://letsencrypt.org/) public certificate authority. These certificates are useful for TLS ingress from web browsers. `Autocert` is different. It's purpose-built to manage certificates issued by your own private CA to support the use of mTLS for internal communication (e.g., service-to-service).
### Questions
#### Doesn't kubernetes already ship with a certificate authority?
#### How is this different than [`cert-manager`](https://github.com/jetstack/cert-manager)
Yes, actually it can have [a bunch of them](https://jvns.ca/blog/2017/08/05/how-kubernetes-certificates-work/) for different sorts of control plane communication.
#### Doesn't kubernetes already ship with a certificate authority?
Wait, no, _actually_
it doesn't _ship_ with _any_ CA. It's complicated. Kubernetes doesn't come with a CA, it has integration points that allow you to use any CA (e.g., [Kubernetes the hard way](https://github.com/kelseyhightower/kubernetes-the-hard-way) [uses CFSSL](https://github.com/kelseyhightower/kubernetes-the-hard-way/blob/2983b28f13b294c6422a5600bb6f14142f5e7a26/docs/02-certificate-authority.md). You could use [`step certificates`](https://github.com/smallstep/certificates), which `autocert` is based on, instead.
Yes, but it's designed for use by the kubernetes control plane rather than by your data plane services. You could use the kubernetes CA to issue certificates for data plane communication, but it's probably not a good idea.
In any case, none of these CAs are meant for issuing certificates to your workloads for service-to-service communication. Rather, they're meant to secure communication between various control plane components. You could use them for your data plane, but it's probably not a good idea.
#### Why not use kubernetes CSR resources for this?
#### What permissions does `autocert` require in my cluster and why?
By default we ask for the narrowest permissions we can: the ability to create and delete secrets cluster-wide. You can [check out our RBAC config here](install/03-rbac.yaml).
We need these permissions in order to transmit one-time tokens to workloads using secrets, and to clean up afterwards. We'd love to scope these permissions down further if anyone has any ideas.
#### Why does the mutating webhook have to create secrets?
It's harder and less secure.
The `autocert` admission webhook needs to securely transmit one-time bootstrap tokens to containers. This could be accomplished without using secrets by simply patching a token directly into the pod's environment via the admission webhook response. Unfortunately, the kubernetes API server does not authenticate itself to admission webhooks by default, and configuring it to do so [requires passing a custom config file](https://kubernetes.io/docs/reference/access-authn-authz/extensible-admission-controllers/#authenticate-apiservers) at apiserver startup. This isn't an option for everyone (e.g., on GKE) so we opted not to rely on it.
Since our webhook can't tell who is calling it, including bootstrap tokens in patch responses would be dangerous. By using secrets an attacker can still trick us into generating superflous bootstrap tokens, but they'd also need read access to cluster secrets to do anything with them.
Hopefully this story will improve with time.
#### Why not use kubernetes service accounts instead of bootstrap tokens?
#### Why does the mutating webhook have to create secrets / need cluster role bindings?
Great idea! This should be pretty easy to add. However, existing service accounts are [somewhat broken](https://github.com/kubernetes/community/pull/1460) for this use case. The upcoming [TokenRequest API](https://github.com/kubernetes/kubernetes/issues/58790) should fix most of these issues.
#### Why do I have to tell you the name to put in a certificate? Why can't you automatically bind service names?
TODO: Link to issue for people who want this.
#### Too. many. containers. Why do you need to install an init container and sidecar?
We don't. Your containers can generate key pairs, exchange them for certificates, and manage renewals themselves. This is pretty easy if you [install `step`](https://github.com/smallstep/cli#installing) in your containers, or integrate with our [golang SDK](https://godoc.org/github.com/smallstep/certificates/ca). To support this we'd need to add the option to inject a bootstrap token without injecting these containers.
#### What are `autocert` certificates good for?
TODO: Link to issue for people who want this.
Autocert certificates let you secure your data plane (service-to-service) communication using mutual TLS (mTLS). Services and proxies can limit access to clients that also have a certificate issued by your certificate authority (CA). Servers can identify which client is connecting improving visibility and enabling granular access control.
That said, the init container and sidecar are both super lightweight.
Once certificates are issued you can use mTLS to secure communication in to, out of, and between kubernetes clusters. Services can use mTLS to only allow connections from clients that have their own certificate issued from your CA.
#### Why are keys and certificates managed via volume mounts? Why not use a secret or some custom resource?
It's like your own Let's Encrypt, but you control who gets a certificate.
Because, by default, kubernetes secrets are stored in plaintext in `etcd` and might even be transmitted unencrypted across the network. Even if secrets were properly encrypted, transmitting a private key across the network violates PKI best practices. Key pairs should always be generated where they're used, and private keys should never be shared with anyone but their owners.
#### How is this different than a service mesh?
That said, there are use cases where a certificate mounted in a secret resource is desirable (e.g., for use with a kubernetes `Ingress`). We may add support for this in the future. However, we think the current method is easier and better.
Certificate management is a necessary building block for any service mesh that uses mutual TLS for authenticated encryption (e.g., istio, linkerd, consul connect). Typically, service mesh systems will provide their own certificate management solution. However, these systems
TODO: Link to issue for people who want this.
#### What about DaemonSets, ReplicaSets, StatefulSets, and all the other things that might need certificates?
#### Why not use kubernetes CSR resources for this?
It's harder and less secure. If any good and simple design exists for securely automating CSR approval using this resource we'd love to see it!
#### Why do I have to tell you the name to put in a certificate? Why can't you automatically bind service names?
Mostly because monitoring the API server to figure out which services are associated with which workloads is complicated and somewhat magical. And it might not be what you want.
That said, we're not totally opposed to this idea if anyone has strong feels and a good design.
...?
#### What sorts of keys are issued and how often are certificates rotated?
`Autocert` builds on `step certificates` which issues ECDSA certificates using the P256 curve with ECDSA-SHA256 signatures by default. If this is all Greek to you, rest assured these are safe, sane, and modern defaults that are suitable for the vast majority of environments.
Use hostnames. Must be global. Everyone who connects to the service using mTLS must use the same hostname. For internal communication it's easy enough to use the FQDN of a service. For stuff you expose publicly you'll need to manage DNS yourself...
In any case, the critical invariant is: ...
Diagram here?
#### Cleaning up one-time token secrets
```
for ns in $(kubectl get namespace --selector autocert.step.sm=enabled -o jsonpath='{$.items[*].metadata.name}'); do
Mike Malone
5 years ago