Skip to content

Latest commit

 

History

History
390 lines (256 loc) · 12.8 KB

README.md

File metadata and controls

390 lines (256 loc) · 12.8 KB
HAProxy Logo

Ansible Role - HAProxy Community (with ACME, GeoIP and some WAF-Features)

Role to deploy HAProxy (Focus on the Community Version)

I think the frontend => route => backend abstraction implemented by this Role is very nice to work with. Please give me some Feedback!

Buy me a coffee

Molecule Test Status YamlLint Test Status PyLint Test Status Ansible-Lint Test Status Ansible Galaxy

Molecule Logs: Short, Full

Tested:

  • Debian 12

Install

# latest
ansible-galaxy role install git+https://github.com/ansibleguy/infra_haproxy

# from galaxy
ansible-galaxy install ansibleguy.infra_haproxy

# or to custom role-path
ansible-galaxy install ansibleguy.infra_haproxy --roles-path ./roles

Roadmap

  • Security
    • Basic rate limit (GET/HEAD and POST/PUT/DELETE separated)
    • Generic client fingerprint
  • 'Interface' for Dict to Map-File translation/creation
  • Option to easily Download & Integrate IPLists (like Tor Exit nodes)
  • Easy way to override the default error-files

Advertisement

  • Need professional support using Ansible or HAProxy? Contact us:

    E-Mail: [email protected]

    Tel: +43 3115 40 900 0

    Web: EN | DE

    Language: German or English

  • You want a simple Ansible GUI?

    Check-out this Ansible WebUI


Usage

Examples

Here some detailed config examples and their results:

Config

Minimal example

haproxy:
  acme:
    enable: true
    email: '[email protected]'

  frontends:
    fe_web:
      bind: ['[::]:80 v4v6', '[::]:443 v4v6 ssl']
      acme:
        enable: true

      routes:
        be_intern:
          domains: ['app.template.ansibleguy.net']

      default_backend: 'be_fallback'

  backends:
    be_intern:
      servers:
        - 'srv-1 192.168.10.11:80'
        - 'srv-2 192.168.10.12:80'

    be_fallback:
      lines: 'http-request redirect code 302 location https://github.com/ansibleguy'

Define the config as needed:

haproxy:
  version: '2.8'
  acme:
    enable: true
    email: '[email protected]'

  # FRONTENDS
  frontends:
    fe_web:
      bind: ['[::]:80 v4v6', '[::]:443 v4v6 ssl']
      acme:
        enable: true
        domains: ['app.template.ansibleguy.net']  # domains from routes will also be added

      routes:
        be_app01:
          domains: ['app01.template.ansibleguy.net', 'hello.template.ansibleguy.net']

      # define raw config sections/lines to add
      lines:
        section1:
          - ...

      default_backend: 'be_fallback'

    fe_dbs:
      mode: 'tcp'
      default_backend: 'be_db'

    fe_restricted:
      bind: ['[::]:8080 v4v6', '[::]:8443 v4v6 ssl crt /etc/myapp/mycert.pem']

      geoip:
        enable: true

      security:
        restrict_methods: true
        allow_only_methods: ['HEAD', 'GET', 'POST']
        fingerprint_ssl: true  # create and log the JA3 SSL-fingerprint of clients
        
        # very basic filtering of bad bots based on user-agent matching
        block_script_bots: true
        block_bad_crawler_bots: true

      routes:
        be_app02:
          filter_country: ['AT', 'DE', 'CH']
          # filter_ip: ['10.0.0.0/8']
          domains: ['app01.template.ansibleguy.net', 'hello.template.ansibleguy.net']

      # define raw config sections/lines to add
      lines:
        section1:
          - ...

      default_backend: 'be_fallback'

  # BACKENDS
  backends:
    be_app01:
      servers:
        - 'app01-1 10.0.1.1:80'
        - 'app01-2 10.0.1.2:80'

      check_uri: '/health'
      check_expect: 'status 200'

    be_app02:
      security:
        # very basic filtering of bad bots based on user-agent matching
        block_script_bots: true
        block_bad_crawler_bots: true

      ssl: true
      ssl_verify: 'none'  # default; example: 'required ca-file /etc/ssl/certs/my_ca.crt verifyhost host01.intern'
      servers:
        - 'app02-1 10.0.1.1:443'
        - 'app02-2 10.0.1.2:443'

    be_db:
      mode: 'tcp'
      balance: 'roundrobin'
          
      # define raw config sections/lines to add
      lines:
        section1:
          - 'option mysql-check user haproxy_check'

      servers:
        - 'mysql-1 10.0.0.1:3306'
        - 'mysql-2 10.0.0.2:3306'

    be_fallback:
      lines:
        default: 'http-request redirect code 302 location https://github.com/ansibleguy'

  # GENERAL
  stats:
    enable: true  # enable stats http listener
    bind: '127.0.0.1:8404'  # default

  geoip:
    enable: true
    provider: 'ipinfo'  # or 'maxmind'
    token: '<YOUR-TOKEN>'

  # define globals/defaults as key/value pairs (multi-value lists usable)
  global:
    ca-base: '/etc/ssl/certs'

  defaults:
    mode: 'http'
    'timeout connect': 3000
    'timeout server': 5000
    'timeout client': 5000

You might want to use 'ansible-vault' to encrypt your passwords:

ansible-vault encrypt_string

Functionality

  • Package installation

  • Configuration

    • Default config:

      • Globals/Defaults - as seen in default installations
    • Default opt-ins:

      • Frontend
        • HTTP mode
          • Redirect non SSL traffic to SSL
          • Logging User-Agent
          • Setting basic security-headers
          • Blocking TRACE & CONNECT methods
    • Default opt-outs:

      • Stats http listener

      • Frontend

      • Backend

        • Sticky sessions
        • Blocking TRACE & CONNECT methods

Info

  • Note: this role currently only supports debian-based systems

  • Note: Most of the role's functionality can be opted in or out.

    For all available options - see the default-config located in the main defaults-file!

  • Warning: Not every setting/variable you provide will be checked for validity. Bad config might break the role!

  • Info: You can easily filter access to backends by using the filter and filter_not settings:

    filter_ip, filter_not_ip, filter_country, filter_not_country, filter_asn, filter_not_asn

  • Info: A very basic user-agent based Script- & Bad-Crawler-Bot blocking can be activated for frontends and backends. Check out the defaults for the list of bots that are blocked.

  • Info: You can easily restrict the HTTP methods allowed on a specific frontend or backend by setting security.restrict_methods to true and specifying security.allow_only_methods

  • Info: Check out the Fingerprinting Docs for detailed information on how you might want to track clients.

  • Info: If you are using Graylog Server to gather and analyze your logs - make sure to split your HAProxy logs into fields using pipeline rules. Example: HAProxy Community - Graylog Pipeline Rule

  • Tip: You can increase the amount of available track-sc's by setting the global setting tune.stick-counters. This can be useful in environments with complex rate-limit setups.

  • Note: You can define the basic_auth.users in the backend-section to enforce a password-prompt before accessing your application. OAuth-Proxy support will be added later on.

GeoIP

  • Warning: If you use the auto-provisioned GeoIP databases - make sure your product follows their license agreement:

    • IPinfo: Information, CC4 License (allows for commercial usage - you need to add an attribution)

      Attribution: <p>IP address data powered by <a href="https://ipinfo.io">IPinfo</a></p>

    • MaxMind: Information, EULA (allows for limited commercial usage - you need to add an attribution)

      Attribution: This product includes GeoLite2 data created by MaxMind, available from <a href="https://www.maxmind.com">https://www.maxmind.com</a>.

  • Info: For GeoIP Tokens you will have to create a free account:

  • Info: If you want to self-manage the GeoIP-databases (not recommended) - the role will assume they are placed at /var/local/lib/geoip and be named asn.mmdb & country.mmdb.

  • Info: You can test the GeoIP Lookup Microservice manually by using curl: curl 'http://127.0.0.1:10069/?lookup=country&ip=1.1.1.1'

WAF

  • Note: The WAF/security feature-set this role provides does not come lose to the one available in HAProxy Enterprise by default. If you have the money - go for it.

  • Tip: If you are using security.flag_bots you can use this basic boolean flag to harden rules for possible bots.

    Examples:

    • Lower rate-limit for bots: http-request deny deny_status 429 if !{ var(txn.bot) -m int 0 } { sc_http_req_rate(0) gt 50 }

    • Hard deny bots to register accounts: http-request deny deny_status 400 if !{ var(txn.bot) -m int 0 } { method POST } { path_sub -m str -i /register/ }

    • Pass the flag to your application to show a pretty error: http-request add-header X-Bot %[var(txn.bot)]

  • Note: If you want to use security.block_script_kiddies make sure you check out the block-list in the defaults and add excludes as needed.

TCP

  • Info: If you want to capture data dynamically, you can use tcp-request content capture.

    You have to enable the logging of captured data manually by modifying the log-format: {% raw %}<DEFAULT LOG FORMAT HERE> {%[capture.req.hdr(0)]|%[capture.req.hdr(1)]}{% endraw %}

    This can be used to log the SNI or GeoIP information.


Execution

Run the playbook:

ansible-playbook -K -D -i inventory/hosts.yml playbook.yml

There are also some useful tags available:

  • install
  • config => only update config and ssl certs
  • ssl or acme
  • geoip
  • lua

To debug errors - you can set the 'debug' variable at runtime:

ansible-playbook -K -D -i inventory/hosts.yml playbook.yml -e debug=yes