This is a Bitwarden server API implementation written in Rust compatible with upstream Bitwarden clients*, perfect for self-hosted deployment where running the official resource-heavy service might not be ideal.

---

Image is based on Rust implementation of Bitwarden API.

*Note, that this project is not associated with the Bitwarden project nor 8bit Solutions LLC.

---

Table of contents

• Features
• Missing features
• Docker image usage
  • Starting a container
  • Updating the bitwarden image
• Configuring bitwarden service
  • Disable registration of new users
  • Disable invitations
  • Configure server administrator
  • Enabling HTTPS
  • Enabling WebSocket notifications
  • Enabling U2F authentication
  • Enabling YubiKey OTP authentication
  • Changing persistent data location
    • /data prefix:
    • database name and location
    • attachments location
    • icons cache
  • Changing the API request size limit
  • Changing the number of workers
  • SMTP configuration
  • Password hint display
  • Disabling or overriding the Vault interface hosting
  • Other configuration
  • Fail2Ban Setup
    • Logging Failed Login Attempts to Syslog
    • Fail2Ban Filter
    • Fail2Ban Jail
    • Testing Fail2Ban
• Building your own image
• Building binary
• Available packages
  • Arch Linux
• Kubernetes deployment
• Backing up your vault
  • 1. the sqlite3 database
  • 2. the attachments folder
  • 3. the key files
  • 4. Icon Cache
• Running the server with non-root user
• Differences from upstream API implementation
  • Changing user email
  • Creating organization
  • Inviting users into organization
  • Running on unencrypted connection
• Get in touch

Features

Basically full implementation of Bitwarden API is provided including:

• Basic single user functionality
• Organizations support
• Attachments
• Vault API support
• Serving the static files for Vault interface
• Website icons API
• Authenticator and U2F support
• YubiKey OTP

Missing features

• Email confirmation
• Other two-factor systems:
  • Duo
  • Email codes

Docker image usage

Starting a container

The persistent data is stored under /data inside the container, so the only requirement for persistent deployment using Docker is to mount persistent volume at the path:

docker run -d --name bitwarden -v /bw-data/:/data/ -p 80:80 mprasil/bitwarden:latest

This will preserve any persistent data under /bw-data/, you can adapt the path to whatever suits you.

The service will be exposed on port 80.

Updating the bitwarden image

Updating is straightforward, you just make sure to preserve the mounted volume. If you used the bind-mounted path as in the example above, you just need to pull the latest image, stop and rm the current container and then start a new one the same way as before:

# Pull the latest version
docker pull mprasil/bitwarden:latest

# Stop and remove the old container
docker stop bitwarden
docker rm bitwarden

# Start new container with the data mounted
docker run -d --name bitwarden -v /bw-data/:/data/ -p 80:80 mprasil/bitwarden:latest

Then visit http://localhost:80

In case you didn't bind mount the volume for persistent data, you need an intermediate step where you preserve the data with an intermediate container:

# Pull the latest version
docker pull mprasil/bitwarden:latest

# Create intermediate container to preserve data
docker run --volumes-from bitwarden --name bitwarden_data busybox true

# Stop and remove the old container
docker stop bitwarden
docker rm bitwarden

# Start new container with the data mounted
docker run -d --volumes-from bitwarden_data --name bitwarden -p 80:80 mprasil/bitwarden:latest

# Optionally remove the intermediate container
docker rm bitwarden_data

# Alternatively you can keep data container around for future updates in which case you can skip last step.

Configuring bitwarden service

Disable registration of new users

By default new users can register, if you want to disable that, set the SIGNUPS_ALLOWED env variable to false:

docker run -d --name bitwarden \
  -e SIGNUPS_ALLOWED=false \
  -v /bw-data/:/data/ \
  -p 80:80 \
  mprasil/bitwarden:latest

Note: While users can't register on their own, they can still be invited by already registered users. Read below if you also want to disable that.

Disable invitations

Even when registration is disabled, organization administrators or owners can invite users to join organization. This won't send email invitation to the users, but after they are invited, they can register with the invited email even if SIGNUPS_ALLOWED is actually set to false. You can disable this functionality completely by setting INVITATIONS_ALLOWED env variable to false:

docker run -d --name bitwarden \
  -e SIGNUPS_ALLOWED=false \
  -e INVITATIONS_ALLOWED=false \
  -v /bw-data/:/data/ \
  -p 80:80 \
  mprasil/bitwarden:latest

Configure server administrator

Warning: Never use your regular account for the admin functionality. This is a bit of a hack using the Vault interface for something it's not intended to do and it breaks any other functionality for the account. Please set up and use separate account just for this functionality.

You can configure one email account to be server administrator via the SERVER_ADMIN_EMAIL environment variable:

docker run -d --name bitwarden \
  -e SERVER_ADMIN_EMAIL=admin@example.com \
  -v /bw-data/:/data/ \
  -p 80:80 \
  mprasil/bitwarden:latest

This will give the user extra functionality and privileges to manage users on the server. In the Vault, the user will see a special (virtual) organization called bitwarden_rs. This organization doesn't actually exist and can't be used for most things. (can't have collections or ciphers) Instead it just contains all the users registered on the server. Deleting users from this organization will actually completely delete the user from the server. Inviting users into this organization will just invite the user so they are able to register, but will not grant any organization membership. (unlike inviting user to regular organization)

You can think of the bitwarden_rs organization as sort of Admin interface to manage users on the server. Keep in mind that deleting user this way removes the user permanently without any way to restore the deleted data just as if user deleted their own account.

Enabling HTTPS

To enable HTTPS, you need to configure the ROCKET_TLS.

The values to the option must follow the format:

ROCKET_TLS={certs="/path/to/certs.pem",key="/path/to/key.pem"}

Where:

• certs: a path to a certificate chain in PEM format
• key: a path to a private key file in PEM format for the certificate in certs

docker run -d --name bitwarden \
  -e ROCKET_TLS='{certs="/ssl/certs.pem",key="/ssl/key.pem"}' \
  -v /ssl/keys/:/ssl/ \
  -v /bw-data/:/data/ \
  -p 443:80 \
  mprasil/bitwarden:latest

Note that you need to mount ssl files and you need to forward appropriate port.

Due to what is likely a certificate validation bug in Android, you need to make sure that your certificate includes the full chain of trust. In the case of certbot, this means using fullchain.pem instead of cert.pem.

Softwares used for getting certs are often using symlinks. If that is the case, both locations need to be accessible to the docker container.

Example: certbot will create a folder that contains the needed fullchain.pem and privkey.pem files in /etc/letsencrypt/live/mydomain/

These files are symlinked to ../../archive/mydomain/privkey.pem

So to use from bitwarden container:

docker run -d --name bitwarden \
  -e ROCKET_TLS='{certs="/ssl/live/mydomain/fullchain.pem",key="/ssl/live/mydomain/privkey.pem"}' \
  -v /etc/letsencrypt/:/ssl/ \
  -v /bw-data/:/data/ \
  -p 443:80 \
  mprasil/bitwarden:latest

Enabling WebSocket notifications

Important: This does not apply to the mobile clients, which use push notifications.

To enable WebSockets notifications, an external reverse proxy is necessary, and it must be configured to do the following:

• Route the /notifications/hub endpoint to the WebSocket server, by default at port 3012, making sure to pass the Connection and Upgrade headers. (Note the port can be changed with WEBSOCKET_PORT variable)
• Route everything else, including /notifications/hub/negotiate, to the standard Rocket server, by default at port 80.
• If using Docker, you may need to map both ports with the -p flag

Example configurations are included in the PROXY.md file.

Then you need to enable WebSockets negotiation on the bitwarden_rs side by setting the WEBSOCKET_ENABLED variable to true:

docker run -d --name bitwarden \
  -e WEBSOCKET_ENABLED=true \
  -v /bw-data/:/data/ \
  -p 80:80 \
  -p 3012:3012 \
  mprasil/bitwarden:latest

Note: The reason for this workaround is the lack of support for WebSockets from Rocket (though it's a planned feature), which forces us to launch a secondary server on a separate port.

Enabling U2F authentication

To enable U2F authentication, you must be serving bitwarden_rs from an HTTPS domain with a valid certificate (Either using the included HTTPS options or with a reverse proxy). We recommend using a free certificate from Let's Encrypt.

After that, you need to set the DOMAIN environment variable to the same address from where bitwarden_rs is being served:

docker run -d --name bitwarden \
  -e DOMAIN=https://bw.domain.tld \
  -v /bw-data/:/data/ \
  -p 80:80 \
  mprasil/bitwarden:latest

Note that the value has to include the https:// and it may include a port at the end (in the format of https://bw.domain.tld:port) when not using 443.

Enabling YubiKey OTP authentication

To enable YubiKey authentication, you must set the YUBICO_CLIENT_ID and YUBICO_SECRET_KEY env variables.

If YUBICO_SERVER is not specified, it will use the default YubiCloud servers. You can generate YUBICO_CLIENT_ID and YUBICO_SECRET_KEY for the default YubiCloud here.

Note: In order to generate API keys or use a YubiKey with an OTP server, it must be registered. After configuring your key in the YubiKey Personalization Tool, you can register it with the default servers here.

docker run -d --name bitwarden \
  -e YUBICO_CLIENT_ID=12345 \
  -e YUBICO_SECRET_KEY=ABCDEABCDEABCDEABCDE= \
  -v /bw-data/:/data/ \
  -p 80:80 \
  mprasil/bitwarden:latest

Changing persistent data location

/data prefix:

By default all persistent data is saved under /data, you can override this path by setting the DATA_FOLDER env variable:

docker run -d --name bitwarden \
  -e DATA_FOLDER=/persistent \
  -v /bw-data/:/persistent/ \
  -p 80:80 \
  mprasil/bitwarden:latest

Notice, that you need to adapt your volume mount accordingly.

database name and location

Default is $DATA_FOLDER/db.sqlite3, you can change the path specifically for database using DATABASE_URL variable:

docker run -d --name bitwarden \
  -e DATABASE_URL=/database/bitwarden.sqlite3 \
  -v /bw-data/:/data/ \
  -v /bw-database/:/database/ \
  -p 80:80 \
  mprasil/bitwarden:latest

Note, that you need to remember to mount the volume for both database and other persistent data if they are different.

attachments location

Default is $DATA_FOLDER/attachments, you can change the path using ATTACHMENTS_FOLDER variable:

docker run -d --name bitwarden \
  -e ATTACHMENTS_FOLDER=/attachments \
  -v /bw-data/:/data/ \
  -v /bw-attachments/:/attachments/ \
  -p 80:80 \
  mprasil/bitwarden:latest

Note, that you need to remember to mount the volume for both attachments and other persistent data if they are different.

icons cache

Default is $DATA_FOLDER/icon_cache, you can change the path using ICON_CACHE_FOLDER variable:

docker run -d --name bitwarden \
  -e ICON_CACHE_FOLDER=/icon_cache \
  -v /bw-data/:/data/ \
  -v /icon_cache/ \
  -p 80:80 \
  mprasil/bitwarden:latest

Note, that in the above example we don't mount the volume locally, which means it won't be persisted during the upgrade unless you use intermediate data container using --volumes-from. This will impact performance as bitwarden will have to re-download the icons on restart, but might save you from having stale icons in cache as they are not automatically cleaned.

Changing the API request size limit

By default the API calls are limited to 10MB. This should be sufficient for most cases, however if you want to support large imports, this might be limiting you. On the other hand you might want to limit the request size to something smaller than that to prevent API abuse and possible DOS attack, especially if running with limited resources.

To set the limit, you can use the ROCKET_LIMITS variable. Example here shows 10MB limit for posted json in the body (this is the default):

docker run -d --name bitwarden \
  -e ROCKET_LIMITS={json=10485760} \
  -v /bw-data/:/data/ \
  -p 80:80 \
  mprasil/bitwarden:latest

Changing the number of workers

When you run bitwarden_rs, it spawns 2 * <number of cpu cores> workers to handle requests. On some systems this might lead to low number of workers and hence slow performance, so the default in the docker image is changed to spawn 10 threads. You can override this setting to increase or decrease the number of workers by setting the ROCKET_WORKERS variable.

In the example below, we're starting with 20 workers:

docker run -d --name bitwarden \
  -e ROCKET_WORKERS=20 \
  -v /bw-data/:/data/ \
  -p 80:80 \
  mprasil/bitwarden:latest

SMTP configuration

You can configure bitwarden_rs to send emails via a SMTP agent:

docker run -d --name bitwarden \
  -e SMTP_HOST=<smtp.domain.tld> \
  -e SMTP_FROM=<bitwarden@domain.tld> \
  -e SMTP_PORT=587 \
  -e SMTP_SSL=true \
  -e SMTP_USERNAME=<username> \
  -e SMTP_PASSWORD=<password> \
  -v /bw-data/:/data/ \
  -p 80:80 \
  mprasil/bitwarden:latest

When SMTP_SSL is set to true(this is the default), only TLSv1.1 and TLSv1.2 protocols will be accepted and SMTP_PORT will default to 587. If set to false, SMTP_PORT will default to 25 and the connection won't be encrypted. This can be very insecure, use this setting only if you know what you're doing.

Password hint display

Usually, password hints are sent by email. But as bitwarden_rs is made with small or personal deployment in mind, hints are also available from the password hint page, so you don't have to configure an email service. If you want to disable this feature, you can use the SHOW_PASSWORD_HINT variable:

docker run -d --name bitwarden \
  -e SHOW_PASSWORD_HINT=false \
  -v /bw-data/:/data/ \
  -p 80:80 \
  mprasil/bitwarden:latest

Disabling or overriding the Vault interface hosting

As a convenience bitwarden_rs image will also host static files for Vault web interface. You can disable this static file hosting completely by setting the WEB_VAULT_ENABLED variable.

docker run -d --name bitwarden \
  -e WEB_VAULT_ENABLED=false \
  -v /bw-data/:/data/ \
  -p 80:80 \
  mprasil/bitwarden:latest

Alternatively you can override the Vault files and provide your own static files to host. You can do that by mounting a path with your files over the /web-vault directory in the container. Just make sure the directory contains at least index.html file.

docker run -d --name bitwarden \
  -v /path/to/static/files_directory:/web-vault \
  -v /bw-data/:/data/ \
  -p 80:80 \
  mprasil/bitwarden:latest

Note that you can also change the path where bitwarden_rs looks for static files by providing the WEB_VAULT_FOLDER environment variable with the path.

Other configuration

Though this is unlikely to be required in small deployment, you can fine-tune some other settings like number of workers using environment variables that are processed by Rocket, please see details in documentation.

Fail2Ban Setup

Bitwarden_rs logs failed login attempts to stdout. We need to set this so the host OS can see these. Then we can setup Fail2Ban.

Logging Failed Login Attempts to Syslog

We need to set the logging driver to syslog so the host OS and Fail2Ban can see them.

If you are using docker commands, you will need to add: --log-driver syslog --log-opt tag=$TAG to your command.

If you are using docker-compose, add this to you yaml file:

  bitwarden:
    logging:
      driver: "syslog"
      options:
        tag: "$TAG"

With the above settings in the docker-compose file. Any failed login attempts will look like this in your syslog file:

MMM DD hh:mm:ss server-hostname $TAG[773]: [YYYY-MM-DD][hh:mm:ss][bitwarden_rs::api::identity][ERROR] Username or password is incorrect. Try again. IP: XXX.XXX.XXX.XXX. Username: email@domain.com.

You can change the '$TAG' to anything you like. Just remember it because it will be in the Fail2Ban filter.

Fail2Ban Filter

Create the filter file

sudo nano /etc/fail2ban/filter.d/bitwarden.conf

And add the following

[INCLUDES]
before = common.conf

[Definition]
_daemon = $TAG
failregex = ^%(__prefix_line)s.*Username or password is incorrect\. Try again\. IP: <HOST>\. Username:.*$
ignoreregex =

Dont forget to change the '$TAG' to what you set it as from above.

Fail2Ban Jail

Now we need the jail, create the jail file

sudo nano /etc/fail2ban/jail.d/bitwarden.local

and add:

[bitwarden]
enabled = true
port = 80,443,8081
filter = bitwarden
action = iptables-allports[name=bitwarden]
logpath = /var/log/syslog
maxretry = 3
bantime = 14400
findtime = 14400

Feel free to change the options as you see fit.

Testing Fail2Ban

Now just try to login to bitwarden using any email (it doesnt 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:

sudo fail2ban-client unban XX.XX.XX.XX bitwarden

Building your own image

Clone the repository, then from the root of the repository run:

# Build the docker image:
docker build -t bitwarden_rs .

Building binary

For building binary outside the Docker environment and running it locally without docker, please see build instructions.

Available packages

Arch Linux

Bitwarden_rs is already packaged for Archlinux thanks to @mqus. There is an AUR package (optionally with the vault web interface ) available.

Kubernetes deployment

Please check the kubernetes-bitwarden_rs repository for example deployment in Kubernetes. It will setup a fully functional and secure bitwarden_rs application in Kubernetes behind nginx-ingress-controller and AWS ELBv1. It provides little bit more than just simple deployment but you can use all or just part of the manifests depending on your needs and setup.

Backing up your vault

1. the sqlite3 database

The sqlite3 database should be backed up using the proper sqlite3 backup command. This will ensure the database does not become corrupted if the backup happens during a database write.

mkdir $DATA_FOLDER/db-backup
sqlite3 /$DATA_FOLDER/db.sqlite3 ".backup '/$DATA_FOLDER/db-backup/backup.sqlite3'"

This command can be run via a CRON job everyday, however note that it will overwrite the same backup.sqlite3 file each time. This backup file should therefore be saved via incremental backup either using a CRON job command that appends a timestamp or from another backup app such as Duplicati. To restore simply overwrite db.sqlite3 with backup.sqlite3 (while bitwarden_rs is stopped).

Running the above command requires sqlite3 to be installed on the docker host system. You can achieve the same result with a sqlite3 docker container using the following command.

docker run --rm --volumes-from=bitwarden bruceforce/bw_backup /backup.sh

You can also run a container with integrated cron daemon to automatically backup your database. See https://gitlab.com/1O/bitwarden_rs-backup for examples.

2. the attachments folder

By default, this is located in $DATA_FOLDER/attachments

3. the key files

This is optional, these are only used to store tokens of users currently logged in, deleting them would simply log each user out forcing them to log in again. By default, these are located in the $DATA_FOLDER (by default /data in the docker). There are 3 files: rsa_key.der, rsa_key.pem, rsa_key.pub.der.

4. Icon Cache

This is optional, the icon cache can re-download itself however if you have a large cache, it may take a long time. By default it is located in $DATA_FOLDER/icon_cache

Running the server with non-root user

The root user inside the container is already pretty limited in what it can do, so the default setup should be secure enough. However if you wish to go the extra mile to avoid using root even in container, here's how you can do that:

1. Create a data folder that's owned by non-root user, so you can use that user to write persistent data. Get the user id. In linux you can run stat <folder_name> to get/verify the owner ID.
2. When you run the container, you need to provide the user ID as one of the parameters. Note that this needs to be in the numeric form and not the username, because docker would try to find such user-defined inside the image, which would likely not be there or it would have different ID than your local user and hence wouldn't be able to write the persistent data. This can be done with the --user parameter.
3. bitwarden_rs listens on port 80 inside the container by default, this won't work with non-root user, because regular users aren't allowed to open port below 1024. To overcome this, you need to configure server to listen on a different port, you can use ROCKET_PORT to do that.

Here's sample docker run, that uses user with id 1000 and with the port redirection configured, so that inside container the service is listening on port 8080 and docker translates that to external (host) port 80:

docker run -d --name bitwarden \
  --user 1000 \
  -e ROCKET_PORT=8080 \
  -v /bw-data/:/data/ \
  -p 80:8080 \
  mprasil/bitwarden:latest

Differences from upstream API implementation

Changing user email

Because we don't have any SMTP functionality at the moment, there's no way to deliver the verification token when you try to change the email. User just needs to enter any random token to continue and the change will be applied.

Creating organization

We use upstream Vault interface directly without any (significant) changes, this is why user is presented with paid options when creating organization. To create an organization, just use the free option, none of the limits apply when using bitwarden_rs as back-end API and after the organization is created it should behave like Enterprise organization.

Inviting users into organization

The invited users won't get the invitation email, instead all already registered users will appear in the interface as if they already accepted the invitation. Organization admin then just needs to confirm them to be proper Organization members and to give them access to the shared secrets.

Invited users, that aren't registered yet will show up in the Organization admin interface as "Invited". At the same time an invitation record is created that allows the users to register even if user registration is disabled. (unless you disable this functionality) They will automatically become "Accepted" once they register. From there Organization admin can confirm them to give them access to Organization.

Running on unencrypted connection

It is strongly recommended to run bitwarden_rs service over HTTPS. However the server itself while supporting it does not strictly require such setup. This makes it a bit easier to spin up the service in cases where you can generally trust the connection (internal and secure network, access over VPN,..) or when you want to put the service behind HTTP proxy, that will do the encryption on the proxy end.

Running over HTTP is still reasonably secure provided you use really strong master password and that you avoid using web Vault over connection that is vulnerable to MITM attacks where attacker could inject javascript into your interface. However some forms of 2FA might not work in this setup and Vault doesn't work in this configuration in Chrome.

Get in touch

To ask an question, raising an issue is fine, also please report any bugs spotted here.

If you prefer to chat, we're usually hanging around at #bitwarden_rs:matrix.org room on Matrix. Feel free to join us!