deSEC Stack

This is a docker-compose application providing the basic stack for deSEC name services. It consists of

• nslord: Eventually authoritative DNS server (PowerDNS). DNSSEC keying material is generated here.
• nsmaster: Stealth authoritative DNS server (PowerDNS). Receives fully signed AXFR zone transfers from nslord. No access to keys.
• api: RESTful API to create deSEC users and domains, see documentation.
• dbapi, dblord, dbmaster: Postgres databases for api and nsmaster, MariaDB database for nslord, respectively.
• www: nginx instance serving static website content and proxying to api
• celery: A shadow instance of the api code for performing asynchronous tasks (email delivery).
• rabbitmq: celery's queue
• memcached: api-wide in-memory cache, currently used to keep API throttling state
• openvpn-server: OpenVPN server used to tunnel replication traffic between this stack and frontend DNS secondaries
• prometheus: Prometheus server for monitoring

Requirements

Although most configuration is contained in this repository, some external dependencies need to be met before the application can be run. Dependencies are:

1. We run this software with the --userland-proxy=false flag of the dockerd daemon, and recommend you do the same.

2. Also, configure certificates for openvpn-server:

   • Get easy-rsa and follow this tutorial.
   • Then, copy ca.crt, server.crt, and server.key to openvpn-server/secrets/.
   • Create a pre-shared secret using openvpn --genkey --secret ta.key inside openvpn-server/secrets/.

   For provisioning a secondary, use the same easy-rsa PKI and create a new client.key and client.crt pair. Transfer these securely onto the secondary, along with ca.crt and ta.key. (You can also create the key on the secondary and only transfer a certificate signing request and the certificate.)

3. Set sensitive information and network topology using environment variables or an .env file. You need (you can use the .env.default file as a template):

   • global
     • DESECSTACK_DOMAIN: domain name under which the entire system will be running. The API will be reachable at https://desec.$DESECSTACK_DOMAIN/api/. For development setup, we recommend using yourname.dedyn.io
     • DESECSTACK_NS: the names of the authoritative name servers, i.e. names pointing to your secondary name servers. Minimum 2.
   • network
     • DESECSTACK_IPV4_REAR_PREFIX16: IPv4 net, size /16, for assignment of internal container IPv4 addresses. NOTE: If you change this in an existing setup, you need to manually update persisted data structures such as the MySQL grant tables! Better don't do it.
     • DESECSTACK_IPV6_SUBNET: IPv6 net, ideally /80 (see below)
     • DESECSTACK_IPV6_ADDRESS: IPv6 address of frontend container, ideally 0642:ac10:0080 in within the above subnet (see below)
     • DESECSTACK_PORT_XFR: Port over which XFRs are performed with secondaries
   • certificates
     • DESECSTACK_WWW_CERTS: ./path/to/certificates for www container. This directory is monitored for changes so that nginx can reload when new keys/certificates are provided. Note: The reload is done any time something changes in the directory. The relevant files are not watched individually.
   • API-related
     • DESECSTACK_API_ADMIN: white-space separated list of Django admin email addresses
     • DESECSTACK_API_AUTHACTION_VALIDITY: number of hours for which authenticated action links (e.g. email verification) should be considered valid (default: 0)
     • DESECSTACK_API_DEBUG: Django debug setting. Must be True (default in docker-compose.dev.yml) or False (default otherwise)
     • DESECSTACK_API_SEPA_CREDITOR_ID: SEPA creditor ID for donations
     • DESECSTACK_API_EMAIL_HOST: when sending email, use this mail server
     • DESECSTACK_API_EMAIL_HOST_USER: username for sending email
     • DESECSTACK_API_EMAIL_HOST_PASSWORD: password for sending email
     • DESECSTACK_API_EMAIL_PORT: port for sending email
     • DESECSTACK_API_SECRETKEY: Django secret
     • DESECSTACK_API_PSL_RESOLVER: Resolver IP address to use for PSL lookups. If empty, the system's default resolver is used.
     • DESECSTACK_DBAPI_PASSWORD_desec: database password for desecapi
     • DESECSTACK_MINIMUM_TTL_DEFAULT: minimum TTL users can set for RRsets. The setting is per domain, and the default defined here is used on domain creation.
   • nslord-related
     • DESECSTACK_DBLORD_PASSWORD_pdns: mysql password for pdns on nslord
     • DESECSTACK_NSLORD_APIKEY: pdns API key on nslord
     • DESECSTACK_NSLORD_CARBONSERVER: pdns carbon-server setting on nslord (optional)
     • DESECSTACK_NSLORD_CARBONOURNAME: pdns carbon-ourname setting on nslord (optional)
     • DESECSTACK_NSLORD_DEFAULT_TTL: TTL to use by default, including for default NS records
   • nsmaster-related
     • DESECSTACK_DBMASTER_PASSWORD_pdns: mysql password for pdns on nsmaster
     • DESECSTACK_NSMASTER_ALSO_NOTIFY: Comma-separated list of additional IP addresses to notify of zone updates
     • DESECSTACK_NSMASTER_APIKEY: pdns API key on nsmaster (required so that we can execute zone deletions on nsmaster, which replicates to the secondaries)
     • DESECSTACK_NSMASTER_CARBONSERVER: pdns carbon-server setting on nsmaster (optional)
     • DESECSTACK_NSMASTER_CARBONOURNAME: pdns carbon-ourname setting on nsmaster (optional)
     • DESECSTACK_NSMASTER_TSIGKEY: Base64-encoded value of the default TSIG key used for talking to external secondaries (algorithm: HMAC-SHA256)
   • monitoring-related
     • DESECSTACK_WATCHDOG_SECONDARIES: space-separated list of secondary hostnames; used to check correct replication of recent DNS changes
     • DESECSTACK_PROMETHEUS_PASSWORD: basic auth password for user prometheus at https://${DESECSTACK_DOMAIN}/prometheus/

How to Run

Development:

$ ./dev

Production:

$ docker-compose build && docker-compose up

Storage

All important data is stored in the databases managed by the db* containers. They use Docker volumes which, by default, reside in /var/lib/docker/volumes/desec-stack_{dbapi_postgres,dblord_mysql,dbmaster_postgres}. This is the location you will want to back up. (Be sure to follow standard MySQL/Postgres backup practices, i.e. make sure things are consistent.)

API Versions and Roadmap

deSEC currently maintains the following API versions:

API Version	URL Prefix	Status	Support Ends
Version 1	/api/v1/	stable	earliest 6 months after v2 is declared stable
Version 2	/api/v2/	unstable

You can find our documentation for all API versions at https://desec.readthedocs.io/. (Select the version of interest in the navigation bar.)

Notes on IPv6

This stack is IPv6-capable. Caveats:

• It is not necessary to start the Docker daemon with --ipv6 or --fixed-cidr-v6. However, it is recommended to run dockerd with --userland-proxy=false to avoid exposing ports on the host IPv6 address through docker-proxy.

• Topology: Assuming 2a01:4f8:a0:12eb::/64 is the host network, and we reserve 2a01:4f8:a0:12eb:deec::/80 for the deSEC stack. Docker has more or less established that IPv6 addresses be composed of the /80 prefix and the container MAC address. We choose the private 06:42:ac MAC prefix, defining a /104 subnet. For the remaining 24 bits of the MAC and IPv6 address, the convention seems to be to use the last 24 bits from the internally assigned IPv4 address. However, the first 8 of these are configurable through the DESECSTACK_IPV4_REAR_PREFIX16 variable. Since we don't want public IPv6 addresses to change if the internal IPv4 net prefix changes, we use 0x10 for bits at position 24--17. We thus arrive at the subnet 2a01:4f8:a0:12eb:deec:642:ac10:0/108 for our public IPv6-enabled Docker containers. The last 16 bits of the IPv6 address we indeed take from the internally assigned IP address. The same procedure is used to set the MAC address of IPv6 containers (they begin with 06:42:ac:10:).

  All other traffic in the /80 subnet is unexpected and therefore rejected. This includes traffic for IPv6 addresses that Docker assigns. (If Docker uses the MAC address for this purpose, the prefix is 02:42:ac which is not part of our public network, so we're safe.)

  Since the above topology is strictly determined by the /80 prefix and the MAC address, we hope that most of the hardcoding can be removed in the future.

• Docker currently exposes IPv6-capable containers fully, without restriction. Therefore, it is necessary to set up a firewall, like (ip6tables)

  -A FORWARD -m state --state RELATED,ESTABLISHED -j ACCEPT
  -A FORWARD -d 2a01:4f8:a0:12eb:deec:642:ac10:0/108 -i eth0 -j ACCEPT
  -A FORWARD -d 2a01:4f8:a0:12eb:deec::/80 -i eth0 -j REJECT --reject-with icmp6-port-unreachable
  

Development: Getting Started Guide

As desec-stack utilizes a number of different technologies and software packages, it requires some effort to setup a stack ready for development. While there are certainly many ways to get started hacking desec-stack, here is one way to do it.

1. Requirements. This guide is intended and tested on Ubuntu 20.20. However, many other Linux distributions will also do fine. For desec-stack, docker and docker-compose are required. Further tools that are required to start hacking are git and curl. Recommended, but not strictly required for desec-stack development is to use certbot along with Let's Encrypt and PyCharm. jq, httpie, libmariadbclient-dev, libpq-dev, python3-dev (>= 3.11) and python3-venv (>= 3.11) are useful if you want to follow this guide. The webapp requires Node.js. To install everything you need for this guide except docker and docker-compose, use

   sudo apt install certbot curl git httpie jq libmariadbclient-dev libpq-dev nodejs npm python3-dev python3-venv libmemcached-dev
   

2. Get the code. Clone this repository to your favorite location.

   git clone git@github.com:desec-io/desec-stack.git
   

3. Obtain Domain Names. To run desec-stack, this guide uses a subdomain of dedyn.io provided by desec.io.

   1. Register a deSEC user account. Check out the instructions in our documentation, in particular the Quickstart section.

   2. Register a dedyn.io subdomain to run your desec-stack on it and set up the IP addresses. For this guide, we assume example.dedyn.io. Register it with:

      DOMAIN=example.dedyn.io
      http POST https://desec.io/api/v1/domains/ Authorization:"Token ${TOKEN}" name:='"'${DOMAIN}'"'
      

      If you receive an answer that is different from status code 201, chances are that the name you chose is already taken by someone else. In that case, repeat the last step with a new name. To setup the necessary IP address records, we create a couple of A and AAAA records that point to localhost. As preparation, create a JSON file dns.json with the following content defining the DNS setup for desec-stack:

      [
          {"type": "A",    "ttl":300, "records": ["127.0.0.1"], "subname": "desec"},
          {"type": "AAAA", "ttl":300, "records": ["::1"],       "subname": "desec"},
          {"type": "A",    "ttl":300, "records": ["127.0.0.1"], "subname": "*.desec"},
          {"type": "AAAA", "ttl":300, "records": ["::1"],       "subname": "*.desec"},
      
          {"type": "A",    "ttl":300, "records": ["127.0.0.1"], "subname": "dedyn"},
          {"type": "AAAA", "ttl":300, "records": ["::1"],       "subname": "dedyn"},
          {"type": "A",    "ttl":300, "records": ["127.0.0.1"], "subname": "*.dedyn"},
          {"type": "AAAA", "ttl":300, "records": ["::1"],       "subname": "*.dedyn"}
      ]
      

      We use the deSEC API to publish the DNS information as defined in dns.json:

      http POST https://desec.io/api/v1/domains/${DOMAIN}/rrsets/ Authorization:"Token ${TOKEN}" < dns.json
      

4. Obtain certificates. desec-stack requires SSL certificates for the above-mentioned desec and dedyn hostnames as well as for various subdomains. (For a complete list, see www/README.md.) While we recommend to obtain signed certificates from Let's Encrypt, it's also possible to let desec-stack generate self-signed certificates on startup by just skipping this step. To use the deSEC certbot hook, first download it to an appropriate location and set up your credentials and domain name.

   mkdir -p ~/bin
   cd ~/bin
   curl https://raw.githubusercontent.com/desec-io/desec-certbot-hook/main/hook.sh > desec_certbot_hook.sh
   touch .dedynauth; chmod 600 .dedynauth
   echo DEDYN_TOKEN=${TOKEN} >> .dedynauth
   echo DEDYN_NAME=${DOMAIN} >> .dedynauth
   chmod +x desec_certbot_hook.sh
   

   Now we use certbot to obtain certificates, using the DNS challenge for authentication.

   certbot \
       --config-dir certbot/config --logs-dir certbot/logs --work-dir certbot/work \
       --manual --text --preferred-challenges dns \
       --manual-auth-hook ~/bin/desec_certbot_hook.sh \
       --manual-cleanup-hook ~/bin/desec_certbot_hook.sh \
       --server https://acme-v02.api.letsencrypt.org/directory \
       -d "*.${DOMAIN}" -d "update.dedyn.${DOMAIN}" -d "update4.dedyn.$DOMAIN" -d "update6.dedyn.$DOMAIN" \
       -d "checkip.dedyn.${DOMAIN}" -d "checkipv4.dedyn.${DOMAIN}" -d "checkipv6.dedyn.${DOMAIN}" \
       certonly
   

   Note that the definition of config, logs and work dir are only necessary if you do not want to run certbot as root. Verifying the DNS challenge takes a while, so allow this command to take a couple of minutes. After successfully retrieving the certificate, you can find them in certbot/config/live/$DOMAIN/. To make them available to desec-stack (in the default location), we copy certificate and keys. In the project root directory,

   mkdir certs
   cd certs
   for SUBNAME in desec www.desec get.desec checkip.dedyn checkipv4.dedyn checkipv6.dedyn dedyn www.dedyn update.dedyn update6.dedyn
   do
       ln -s cer ${SUBNAME}.${DOMAIN}.cer
       ln -s key ${SUBNAME}.${DOMAIN}.key
   done
   
   cp ~/bin/certbot/config/live/${DOMAIN}/fullchain.pem cer
   cp ~/bin/certbot/config/live/${DOMAIN}/privkey.pem key
   

   The last two steps need to be repeated whenever the certificates are renewed. While any location for the certificates is fine, the certs/ folder is configured to be ignored by git so that private keys do not accidentally end up being committed.

5. Configure desec-stack. As docker-compose application, desec-stack is configured by environment variables defined in the .env file in the project root directory. Because it contains sensitive information for each deployment, .env is not part of the repository and ignored by git. However, we ship .env.default and .env.dev with templates for production and development, respectively. .env.dev is almost good enough for a basic development system, so let's use that as a basis:

   sed "s/^DESECSTACK_DOMAIN=.*/DESECSTACK_DOMAIN=${DOMAIN}/" .env.dev > .env
   

   Optionally, edit the file and

   1. configure an email server host name, username, and password to deliver emails can be included in .env. A convenient option is a MailTrap account.
   2. adjust the network prefixes in .env to avoid collisions with other local networks.

   Additionally, the VPN server for the replication network needs to be equipped with a pre-shared key (PSK) and a public key infrastructure (PKI). To generate the PSK, use the openvpn-server container:

    docker-compose build openvpn-server && docker-compose run openvpn-server openvpn --genkey --secret /dev/stdout > openvpn-server/secrets/ta.key
   

   To build the PKI, we recommend easy RSA. Please note that PKI instructions here are for development deployments only! Using this setup for production WILL DEFINITELY result in an INSECURE deployment! To make it available, clone the repository and link to the executable:

    cd openvpn-server/secrets
    git clone https://github.com/OpenVPN/easy-rsa.git
    ln -s easy-rsa/easyrsa3/easyrsa
   

   In order to create a new PKI,

    ./easyrsa init-pki
    ./easyrsa build-ca nopass
   

   To make the new PKI's Certificate Authority available to the OpenVPN server,

    ln -s pki/ca.crt
   

   To issue a certificate for the OpenVPN server, generate a new key pair, a signing request, and sign the certificate.

     ./easyrsa gen-req server nopass
     ./easyrsa sign-req server server  # requires interaction
   

   Make the key and certificate available to OpenVPN server:

    ln -s pki/issued/server.crt
    ln -s pki/private/server.key
   

   As the setup of OpenVPN is completed, return to the project directory:

    cd -
   

6. Install webapp dependencies. To install the dependencies for the web site and GUI, run

   cd webapp/
   npm install
   cd -
   

7. Run desec-stack. To run desec-stack, use

   ./dev
   

   If you run desec-stack for the first time, this will require a couple of downloads and take a while. Once it is up and running, you can query the API home endpoint:

   http GET https://desec.${DOMAIN}/api/v1/
   

   Congratulations, you have desec-stack up and running.

   A convenient way to create a test user account is via

   docker-compose exec api python3 manage.py shell -c 'from desecapi.models import User; User.objects.create_user(email="test@example.com", password="test1234");'
   

   but users can also be created by signing up via the web GUI. The latter, however, requires that you can read email that is sent from your local setup. This can be achieved, e.g., by using mailtrap.io.

   Of course, as this setup is only on your local machine, DNS information will not be published into the public DNS. However, the desec-stack nameserver is available on localhost port 5321. To check if desec-stack is working as expected, you can query the desec-stack nameserver locally for any information that you saved using your API.

   EMAIL=john@example.com
   PASSWORD=insecure
   # Register account (https://desec.readthedocs.io/en/latest/quickstart.html). Hint: In dev mode, the captcha response contains the plaintext challenge.
   TOKEN=$(http POST https://desec.${DOMAIN}/api/v1/auth/login/ email:=\"${EMAIL}\" password:=\"${PASSWORD}\" | jq -r .token)
   http POST https://desec.${DOMAIN}/api/v1/domains/ Authorization:"Token ${TOKEN}" name:='"test.example"'
   http POST https://desec.${DOMAIN}/api/v1/domains/test.example/rrsets/ Authorization:"Token ${TOKEN}" type:=\"A\" ttl:=60 records:='["127.0.0.254"]'
   

   After registering a user with your API, creating a domain and publishing some info to the DNS, use

   dig @localhost -p 5321 test.example 
   

   to see if the nameserver is behaving as expected.

8. (Optional) Configure PyCharm for API Development. As a docker-compose application, desec-stack takes a while to start. Additionally, it is hard to connect a debugger to the docker containers. Our recommended solution is to develop the API using Django tests running outside the docker-compose application. This will dramatically decrease the time required for running the Django tests and enable just-in-time debugging in PyCharm. Also, it will enable you to browse dependencies and code within PyCharm and thus ease debugging.

   1. To get started, we create a virtual python environment that (to some extent) mimics the python environment in the docker container. In the project root,

      cd api
      python3 -m venv venv  # Python >= 3.11
      source venv/bin/activate
      pip install wheel
      pip install -r requirements.txt
      

   2. At this point, Django is ready to run in the virtual environment created above. There are two things to consider when running Django outside the container. First, the environment variables as defined in the .env file need to be made available in the shell. This can be done with

      set -a && source ../.env && set +a
      

      Second, to make the tests run efficiently, a couple of settings are different from the production system: passwords are hashed using a fast (but insecure!) method, rate limits are switched off, and so on. To use the fast settings in your shell, run

      export DJANGO_SETTINGS_MODULE=api.settings_quick_test
      

      Third, the API needs a postgres database to run the tests. To serve as a test database, the dbapi container can be started using a test configuration which exposes the database at 127.0.0.1. In order to let Django know that the database is at 127.0.0.1 instead of the usual dbapi, set an additional environment variable:

      export DESECSTACK_DJANGO_TEST=1
      

      Fourth, run the database:

      docker-compose -f docker-compose.yml -f docker-compose.test-api.yml up -d dbapi
      

      Finally, you can manage Django using the manage.py CLI. As an example, to run the tests, use

      python3 manage.py test
      

   3. Open the project root directory desec-stack in PyCharm and select File › Settings.

      1. In Project: desec-stack › Project Structure, mark the api/ folder as a source folder.
      2. In Project: desec-stack › Project Interpreter, add a new interpreter. Choose "existing environment" and select api/venv/bin/python3 from the project root.
      3. In Languages & Frameworks › Django, enable the Django support and set the Django project root to api/.

   4. From the PyCharm menu, select Run › Edit Configurations and click on "Edit configuration templates"; select the "Django tests" template from the list.

      1. Open the Environment Variables dialog. Copy the contents of the .env file and paste it here.
      2. Add an environment variable with the name DESECSTACK_DJANGO_TEST and the value 1.
      3. Fill the Custom Settings field with the path to the settings_quick_test module.
      4. At the bottom in the "Before launch" sections, add an "External tool" with the following settings:
         • Name: Postgres Test Container
         • Program: docker-compose
         • Arguments: -f docker-compose.yml -f docker-compose.test-api.yml up -d dbapi

   5. To see if the test configuration is working, right-click on the api folder in the project view and select Run Test. (Note that the first attempt may fail in case the dbapi container does not start up fast enough. In that case, just try again.)

   6. To use code inspection, click on Inspect Code… in PyCharm's Code menu and add a local custom scope with the following pattern:

      file:api//*.py&&!file:api/venv//*&&!file:api/manage.py&&!file:api/api/wsgi.py&&!file:api/desecapi/migrations//*
      

   From this point on, you are set up to use most of PyCharm's convenience features.

   1. For PyCharm's Python Console, the environment variables of your .env file and DJANGO_SETTINGS_MODULE=api.settings_quick_test need to be configured in Settings › Build, Execution, Deployment › Console › Django Console. (Note that if you need to work with the database, you need to initialize it first by running all migrations; otherwise, the model tables will be missing from the database.)

9. Code quality. We use Black to ensure formatting consistency and minimal diffs. Before you commit Python code into the api/ directory, please run black api/desecapi/.

Debugging

RabbitMQ

To access message queue information of RabbitMQ, use the RabbitMQ management plugin. First, port 15672 of the RabbitMQ container needs to be exposed (default when using docker-compose.dev.yml). Then, inside the container, create a user that can access the RabbitMQ data:

rabbitmq-plugins enable rabbitmq_management
rabbitmqctl add_user admin admin
rabbitmqctl set_user_tags admin administrator
rabbitmqctl set_permissions admin '.*' '.*' '.*'

Then the web-based management interface will be available at http://localhost:15672 with user admin and password admin.