Fail2Ban Setup - dani-garcia/vaultwarden GitHub Wiki

Setting up Fail2ban will prevent attackers to brute force your vault logins. This is particularly important if your instance is publicly available.

• Pre-requisite
• Installation
  • Debian / Ubuntu / Raspberry Pi OS
  • Fedora / Centos
  • Synology DSM
• Setup for web vault
  • Filter
  • Jail
• Setup for admin page
  • Filter
  • Jail
• Testing Fail2Ban
• SELinux Problems

• Filenames are at the top of each code block.
• From Release 1.5.0, Vaultwarden supports logging to file. Please set this up : Logging
• Try to log to web vault with a false account and check the log files for following format

[YYYY-MM-DD hh:mm:ss][vaultwarden::api::identity][ERROR] Username or password is incorrect. Try again. IP: XXX.XXX.XXX.XXX. Username: [email protected].

sudo apt-get install fail2ban -y

EPEL repository is necessary (CentOS 7)

sudo yum install epel-release
sudo yum install fail2ban -y

With Synology, a bit more work is needed for various reasons. The full solution is pushed with Docker Compose there. The main issues are:

1. The embedded IP ban system does not work for Docker's containers
2. The iptables embedded do no support the REJECT blocktype
3. The Docker GUI does not allow some advanced settings
4. Modifying system configuration is not upgrade-proof

Therefore, we will use Fail2ban in a Docker container. Crazy-max/docker-fail2ban provides a good solution and the Synology's Docker GUI will be ignored. From command line through SSH, here the steps. As a convention volumeX is to be adapted to your Synology's config.

0. Get root

sudo -i

1. Creating persistent folders

mkdir -p /volumeX/docker/fail2ban/action.d/
mkdir -p /volumeX/docker/fail2ban/jail.d/
mkdir -p /volumeX/docker/fail2ban/filter.d/

2. Replace REJECT by DROP blocktype

# /volumeX/docker/fail2ban/action.d/iptables.local

[Init]
blocktype = DROP
[Init?family=inet6]
blocktype = DROP

3. Create docker-compose file

# /volumeX/docker/fail2ban/docker-compose.yml

version: '3'
services:
  fail2ban:
    container_name: fail2ban
    restart: always
    image: crazymax/fail2ban:latest
    environment: 
      - TZ=Europe/Paris
      - F2B_DB_PURGE_AGE=30d
      - F2B_LOG_TARGET=/data/fail2ban.log
      - F2B_LOG_LEVEL=INFO
      - F2B_IPTABLES_CHAIN=INPUT

    volumes:
      - /volumeX/docker/fail2ban:/data
      - /volumeX/docker/vw-data:/vaultwarden:ro

    network_mode: "host"

    privileged: true
    cap_add:
      - NET_ADMIN
      - NET_RAW

4. Start the container using command line

cd /volumeX/docker/fail2ban
docker-compose up -d

You should see the container running in Synology's Docker GUI. You will have to reload after configuring the filters and jails

As a convention, path_f2b means the path needed for Fail2ban to work. This depends on your system. E.g. on Synology, we are talking about /volumeX/docker/fail2ban/ where on some other systems we are talking about /etc/fail2ban/

Create and fill the following file

# path_f2b/filter.d/vaultwarden.local

[INCLUDES]
before = common.conf

[Definition]
failregex = ^.*?Username or password is incorrect\. Try again\. IP: <ADDR>\. Username:.*$
ignoreregex =

Tip: If you get the following error message in fail2ban.log (CentOS 7, Fail2Ban v0.9.7)
fail2ban.filter [5291]: ERROR No 'host' group in '^.*Username or password is incorrect\. Try again\. IP: <ADDR>\. Username:.*$'
Please Use <HOST> instead of <ADDR> in vaultwarden.local

Tip: Cloudflare users, make sure you set your Client IP header to CF-Connecting-IP in admin panel -> advanced settings -> Client IP header, else the clients real IP will not be logged/banned.

Tip: If you see 127.0.0.1 as the IP address of failed logins in vaultwarden.log, then you're probably using a reverse proxy and fail2ban won't work correctly:

[YYYY-MM-DD hh:mm:ss][vaultwarden::api::identity][ERROR] Username or password is incorrect. Try again. IP: 127.0.0.1. Username: [email protected].

To remedy this, forward the true remote address to vaultwarden via the X-Real-IP header. How to do this varies depending on the proxy you use. For example, in Caddy 2.x, when you define the reverse-proxy, define header_up X-Real-IP {remote_host}. See Proxy examples for more info.

Create and fill the following file

# path_f2b/jail.d/vaultwarden.local

[vaultwarden]
enabled = true
port = 80,443,8081
filter = vaultwarden
banaction = %(banaction_allports)s
logpath = /path/to/vaultwarden.log
maxretry = 3
bantime = 14400
findtime = 14400

Docker uses the FORWARD chain instead of the default INPUT chain. If the machine receiving requests is mapping them straight to a Docker container, then chain will need to be set appropriately regardless of what is in the container (reverse proxy, Vaultwarden, etc). The default action is set to action_ (which uses banaction, which we alias to banaction_allports). action_ already takes the chain into account. Thus, simply set the chain. See this similar issue.

chain = FORWARD

Please set the chain to DOCKER-USER

chain = DOCKER-USER

With Fail2Ban v1.1.1.dev1 the default banactions for Debian changed from iptables to nftables (see here). Docker (at least version 25.0.3) on the other hand still uses iptables. Hence, the requests to the Docker containers are not blocked with banaction = %(banaction_allports)s. In this scenario, use

banaction = iptables

Tip: If you are using systemd to manage vaultwarden, you can use systemd-journal for fail2ban:

backend = systemd
filter = vaultwarden[journalmatch='_SYSTEMD_UNIT=your_vaultwarden.service']

Use these instead of logpath = and filter = variables.

NOTE FOR CLOUDFLARE USERS If you use cloudflare proxy, you'll need to add Cloudflare in your actions list, like in this guide

Reload fail2ban for changes to take effect:

sudo systemctl reload fail2ban

Feel free to change the options as you see fit.

If you've enabled the admin console by setting the ADMIN_TOKEN environment variable, you can prevent an attacker from brute-forcing the admin token using Fail2Ban. The process is the same as for the web vault.

Create and fill the following file

# path_f2b/filter.d/vaultwarden-admin.local

[INCLUDES]
before = common.conf

[Definition]
failregex = ^.*Invalid admin token\. IP: <ADDR>.*$
ignoreregex =

Tip: If you get the following error message in fail2ban.log ERROR NOK: ("No 'host' group in '^.*Invalid admin token\\. IP: <ADDR>.*$'")
Please Use <HOST> instead of <ADDR> in vaultwarden-admin.local

Create and fill the following file

# path_f2b/jail.d/vaultwarden-admin.local

[vaultwarden-admin]
enabled = true
port = 80,443
filter = vaultwarden-admin
banaction = %(banaction_allports)s
logpath = /path/to/vaultwarden.log
maxretry = 3
bantime = 14400
findtime = 14400

Note: Docker uses the FORWARD chain instead of the default INPUT chain. Therefore replace the banaction line with the following action when using Docker:

action = iptables-allports[name=vaultwarden-admin, chain=FORWARD]

Tip: If you are using systemd to manage vaultwarden, you can use systemd-journal for fail2ban here as well:

backend = systemd
filter = vaultwarden-admin[journalmatch='_SYSTEMD_UNIT=your_vaultwarden.service']

Use these instead of logpath = and filter = variables.

NOTE FOR CLOUDFLARE USERS If you use cloudflare proxy, you'll need to add Cloudflare in your actions list, like in this guide

Reload fail2ban for changes to take effect:

sudo systemctl reload fail2ban

Now just try to login to vaultwarden using any email (it doesn't have to be a valid email, just an email format) If it works correctly and your IP is banned, you can unban the IP by running:

# With Docker
sudo docker exec -t fail2ban fail2ban-client set vaultwarden unbanip XX.XX.XX.XX
# Without Docker
sudo fail2ban-client set vaultwarden unbanip XX.XX.XX.XX

If Fail2Ban does not appear to be functioning, verify that the path to the Vaultwarden log file is correct. For Docker: If the specified log file is not being generated and/or updated, make sure the EXTENDED_LOGGING env variable is set to true (which is default) and that the path to the log file is the path inside the Docker (when you use /bw-data/:/data/ the log file should be in /data/... to be outside the container).

Also verify that the timezone of the Docker container matches the timezone of the host. Check this by comparing the time shown in the logfile with the host OS time. If they differ, there are various ways to fix this. One option is to start Docker with the option -e "TZ=<timezone>". A list of valid timezones is here (eg. -e "TZ=Australia/Melbourne")

If you are using podman instead of Docker it seems that setting the timezone via -e "TZ=<timezone>" does not work. This can be solved (when using the alpine image) by following this guide: https://wiki.alpinelinux.org/wiki/Setting_the_timezone.

When you are using SELinux it is possible that SELinux hinders fail2ban to read the logs. If so, follow these steps: sudo tail /var/log/audit/audit.log. There you should see something along the lines of this (of course the actual audit ID will vary in your case):

type=AVC msg=audit(1571777936.719:2193): avc:  denied  { search } for  pid=5853 comm="fail2ban-server" name="containers" dev="dm-0" ino=1144588 scontext=system_u:system_r:fail2ban_t:s0 tcontext=unconfined_u:object_r:container_var_lib_t:s0 tclass=dir permissive=0

To actually find out the reason you can use grep 'type=AVC msg=audit(1571777936.719:2193)' /var/log/audit/audit.log | audit2why. audit2allow -a will give you specific instructions on how to create a module and allow fail2ban to access these logs. Follow these steps and you're done! fail2ban should now work correctly.