open-vault/website/source/docs/config/index.html.md
2016-04-05 22:18:44 -07:00

20 KiB
Raw Blame History

layout page_title sidebar_current description
docs Server Configuration docs-config Vault server configuration reference.

Server Configuration

Outside of development mode, Vault servers are configured using a file. The format of this file is HCL or JSON. An example configuration is shown below:

backend "consul" {
  address = "127.0.0.1:8500"
  path = "vault"
}

listener "tcp" {
  address = "127.0.0.1:8200"
  tls_disable = 1
}

telemetry {
  statsite_address = "127.0.0.1:8125"
  disable_hostname = true
}

After the configuration is written, use the -config flag with vault server to specify where the configuration is.

Starting with 0.5.2, limited configuration options can be changed on-the-fly by sending a SIGHUP to the server process. These are denoted below.

Reference

  • backend (required) - Configures the storage backend where Vault data is stored. There are multiple options available for storage backends, and they're documented below.

  • ha_backend (optional) - Configures the storage backend where Vault HA coordination will take place. Must be an HA-supporting backend using the configuration options as documented below. If not set, HA will be attempted on the backend given in the backend parameter.

  • listener (required) - Configures how Vault is listening for API requests. "tcp" is currently the only option available. A full reference for the inner syntax is below.

  • disable_cache (optional) - A boolean. If true, this will disable the read cache used by the physical storage subsystem. This will very significantly impact performance.

  • disable_mlock (optional) - A boolean. If true, this will disable the server from executing the mlock syscall to prevent memory from being swapped to disk. This is not recommended in production (see below).

  • telemetry (optional) - Configures the telemetry reporting system (see below).

  • default_lease_ttl (optional) - Configures the default lease duration for tokens and secrets. This is a string value using a suffix, e.g. "720h". Default value is 30 days. This value cannot be larger than max_lease_ttl.

  • max_lease_ttl (optional) - Configures the maximum possible lease duration for tokens and secrets. This is a string value using a suffix, e.g. "720h". Default value is 30 days.

In production it is a risk to run Vault on systems where mlock is unavailable or the setting has been disabled via the disable_mlock. Disabling mlock is not recommended unless the systems running Vault only use encrypted swap or do not use swap at all. Vault only supports memory locking on UNIX-like systems (Linux, FreeBSD, Darwin, etc). Non-UNIX like systems (e.g. Windows, NaCL, Android) lack the primitives to keep a process's entire memory address space from spilling disk and is therefore automatically disabled on unsupported platforms.

On Linux, to give the Vault executable the ability to use the mlock syscall without running the process as root, run:

sudo setcap cap_ipc_lock=+ep $(readlink -f $(which vault))

Listener Reference

For the listener section, the only supported listener currently is "tcp". Regardless of future plans, this is the recommended listener, since it allows for HA mode.

The supported options are:

  • address (optional) - The address to bind to for listening. This defaults to "127.0.0.1:8200".

  • tls_disable (optional) - If true, then TLS will be disabled. This will parse as boolean value, and can be set to "0", "no", "false", "1", "yes", or "true". This is an opt-in; Vault assumes by default that TLS will be used.

  • tls_cert_file (required unless disabled) - The path to the certificate for TLS. This is reloaded via SIGHUP.

  • tls_key_file (required unless disabled) - The path to the private key for the certificate. This is reloaded via SIGHUP.

  • tls_min_version (optional) - (Vault > 0.2) If provided, specifies the minimum supported version of TLS. Accepted values are "tls10", "tls11" or "tls12". This defaults to "tls12". WARNING: TLS 1.1 and lower are generally considered less secure; avoid using these if possible.

Telemetry Reference

For the telemetry section, there is no resource name. All configuration is within the object itself.

  • statsite_address (optional) - An address to a Statsite instances for metrics. This is highly recommended for production usage.

  • statsd_address (optional) - This is the same as statsite_address but for StatsD.

  • disable_hostname (optional) - Whether or not to prepend runtime telemetry with the machines hostname. This is a global option. Defaults to false.

Backend Reference

For the backend section, the supported physical backends are shown below. Vault requires that the backend itself will be responsible for backups, durability, etc.

Please note: The only physical backends actively maintained by HashiCorp are consul, inmem, and file. The other backends are community-derived and community-supported. We include them in the hope that they will be useful to those users that wish to utilize them, but they receive minimal validation and testing from HashiCorp, and HashiCorp staff may not be knowledgeable about the data store being utilized. If you encounter problems with them, we will attempt to help you, but may refer you to the backend author.

  • consul - Store data within Consul. This backend supports HA. It is the most recommended backend for Vault and has been shown to work at high scale under heavy load.

  • etcd - Store data within etcd. This backend supports HA. This is a community-supported backend.

  • zookeeper - Store data within Zookeeper. This backend supports HA. This is a community-supported backend.

  • dynamodb - Store data in a DynamoDB table. This backend supports HA. This is a community-supported backend.

  • s3 - Store data within an S3 bucket S3. This backend does not support HA. This is a community-supported backend.

  • mysql - Store data within MySQL. This backend does not support HA. This is a community-supported backend.

  • postgresql - Store data within PostgreSQL. This backend does not support HA. This is a community-supported backend.

  • inmem - Store data in-memory. This is only really useful for development and experimentation. Data is lost whenever Vault is restarted.

  • file - Store data on the filesystem using a directory structure. This backend does not support HA.

Common Backend Options

All backends support the following options:

  • advertise_addr (optional) - For backends that support HA, this is the address to advertise to other Vault servers in the cluster for request forwarding. As an example, if a cluster contains nodes A, B, and C, node A should set it to the address that B and C should redirect client nodes to when A is the active node and B and C are standby nodes. This may be the same address across nodes if using a load balancer or service discovery. Most HA backends will attempt to determine the advertise address if not provided. This can also be set via the VAULT_ADVERTISE_ADDR environment variable.

Backend Reference: Consul

For Consul, the following options are supported:

  • path (optional) - The path within Consul where data will be stored. Defaults to "vault/".

  • address (optional) - The address of the Consul agent to talk to. Defaults to the local agent address, if available.

  • scheme (optional) - "http" or "https" for talking to Consul.

  • token (optional) - An access token to use to write data to Consul.

  • max_parallel (optional) - The maximum number of connections to Consul; defaults to "128".

  • tls_skip_verify (optional) - If non-empty, then TLS host verification will be disabled for Consul communication. Defaults to false.

The following settings should be set according to your Consul encryption settings:

  • tls_ca_file (optional) - The path to the CA certificate used for Consul communication. Defaults to system bundle if not specified. Set accordingly to the ca_file setting in Consul.

  • tls_cert_file (optional) - The path to the certificate for Consul communication. Set accordingly to the cert_file setting in Consul.

  • tls_key_file (optional) - The path to the private key for Consul communication. Set accordingly to the key_file setting in Consul.

Backend Reference: etcd (Community-Supported)

For etcd, the following options are supported:

  • path (optional) - The path within etcd where data will be stored. Defaults to "vault/".

  • address (optional) - The address(es) of the etcd instance(s) to talk to. Can be comma separated list (protocol://host:port) of many etcd instances. Defaults to "http://localhost:2379" if not specified.

  • sync (optional) - Should we synchronize the list of available etcd servers on startup? This is a string value to allow for auto-sync to be implemented later. It can be set to "0", "no", "n", "false", "1", "yes", "y", or "true". Defaults to on. Set to false if your etcd cluster is behind a proxy server and syncing causes Vault to fail.

  • username (optional) - Username to use when authenticating with the etcd server. May also be specified via the ETCD_USERNAME environment variable.

  • password (optional) - Password to use when authenticating with the etcd server. May also be specified via the ETCD_PASSWORD environment variable.

  • tls_ca_file (optional) - The path to the CA certificate used for etcd communication. Defaults to system bundle if not specified.

  • tls_cert_file (optional) - The path to the certificate for etcd communication.

  • tls_key_file (optional) - The path to the private key for etcd communication.

Backend Reference: Zookeeper (Community-Supported)

For Zookeeper, the following options are supported:

  • path (optional) - The path within Zookeeper where data will be stored. Defaults to "vault/".

  • address (optional) - The address(es) of the Zookeeper instance(s) to talk to. Can be comma separated list (host:port) of many Zookeeper instances. Defaults to "localhost:2181" if not specified.

The following optional settings can be used to configure zNode ACLs:

  • auth_info (optional) - Authentication string in Zookeeper AddAuth format (schema:auth). As an example, digest:UserName:Password could be used to authenticate as user UserName using password Password with the digest mechanism.

  • znode_owner (optional) - If specified, Vault will always set all permissions (CRWDA) to the ACL identified here via the Schema and User parts of the Zookeeper ACL format. The expected format is schema:user-ACL-match. Some examples:

    • digest:UserName:HIDfRvTv623G== - Access for the user UserName with the corresponding digest HIDfRvTv623G==
    • ip:127.0.0.1 - Access from localhost only
    • ip:70.95.0.0/16 - Any host on the 70.95.0.0 network (CIDRs are supported starting from Zookeeper 3.5.0)

If neither of these is set, the backend will not authenticate with Zookeeper and will set the OPEN_ACL_UNSAFE ACL on all nodes. In this scenario, anyone connected to Zookeeper could change Vaults znodes and, potentially, take Vault out of service.

Some sample configurations:

backend "zookeeper" {
  znode_owner = "digest:vaultUser:raxgVAfnDRljZDAcJFxznkZsExs="
  auth_info = "digest:vaultUser:abc"
}

The above configuration causes Vault to set an ACL on all of its zNodes permitting access to vaultUser only. If the digest schema is used, please protect this file as it contains the cleartext password. As per Zookeeper's ACL model, the digest value (in znode_owner) must match the user (in znode_owner).

backend "zookeeper" {
  znode_owner = "ip:127.0.0.1"
}

The above example allows access from localhost only - as this is the ip no auth_info is required since Zookeeper uses the address of the client for the ACL check.

Backend Reference: DynamoDB (Community-Supported)

The DynamoDB backend has the following options:

  • table (optional) - The name of the DynamoDB table to store data in. The default table name is vault-dynamodb-backend. This option can also be provided via the environment variable AWS_DYNAMODB_TABLE. If the specified table does not yet exist, it will be created during initialization.

  • read_capacity (optional) - The read capacity to provision when creating the DynamoDB table. This is the maximum number of reads consumed per second on the table. The default value is 5. This option can also be provided via the environment variable AWS_DYNAMODB_READ_CAPACITY.

  • write_capacity (optional) - The write capacity to provision when creating the DynamoDB table. This is the maximum number of writes performed per second on the table. The default value is 5. This option can also be provided via the environment variable AWS_DYNAMODB_WRITE_CAPACITY.

  • access_key - (required) The AWS access key. It must be provided, but it can also be sourced from the AWS_ACCESS_KEY_ID environment variable.

  • secret_key - (required) The AWS secret key. It must be provided, but it can also be sourced from the AWS_SECRET_ACCESS_KEY environment variable.

  • session_token - (optional) The AWS session token. It can also be sourced from the AWS_SESSION_TOKEN environment variable.

  • endpoint - (optional) An alternative (AWS compatible) DynamoDB endpoint to use. It can also be sourced from the AWS_DYNAMODB_ENDPOINT environment variable.

  • region (optional) - The AWS region. It can be sourced from the AWS_DEFAULT_REGION environment variable and will default to "us-east-1" if not specified.

  • recovery_mode (optional) - When the Vault leader crashes or is killed without being able to shut down properly, no other node can become the new leader because the DynamoDB table still holds the old leader's lock record. To recover from this situation, one can start a single Vault node with this option set to 1 and the node will remove the old lock from DynamoDB. It is important that only one node is running in recovery mode! After this node has become the leader, other nodes can be started with regular configuration. This option can also be provided via the environment variable RECOVERY_MODE.

For more information about the read/write capacity of DynamoDB tables, see the official AWS DynamoDB docs. If you are running your Vault server on an EC2 instance, you can also make use of the EC2 instance profile service to provide the credentials Vault will use to make DynamoDB API calls. Leaving the access_key and secret_key fields empty will cause Vault to attempt to retrieve credentials from the metadata service.

Backend Reference: S3 (Community-Supported)

For S3, the following options are supported:

  • bucket (required) - The name of the S3 bucket to use. It must be provided, but it can also be sourced from the AWS_S3_BUCKET environment variable.

  • access_key - (required) The AWS access key. It must be provided, but it can also be sourced from the AWS_ACCESS_KEY_ID environment variable.

  • secret_key - (required) The AWS secret key. It must be provided, but it can also be sourced from the AWS_SECRET_ACCESS_KEY environment variable.

  • session_token - (optional) The AWS session token. It can also be sourced from the AWS_SESSION_TOKEN environment variable.

  • endpoint - (optional) An alternative (AWS compatible) S3 endpoint to use. It can also be sourced from the AWS_S3_ENDPOINT environment variable.

  • region (optional) - The AWS region. It can be sourced from the AWS_DEFAULT_REGION environment variable and will default to "us-east-1" if not specified.

If you are running your Vault server on an EC2 instance, you can also make use of the EC2 instance profile service to provide the credentials Vault will use to make S3 API calls. Leaving the access_key and secret_key fields empty will cause Vault to attempt to retrieve credentials from the metadata service. You are responsible for ensuring your instance is launched with the appropriate profile enabled. Vault will handle renewing profile credentials as they rotate.

Backend Reference: MySQL (Community-Supported)

The MySQL backend has the following options:

  • username (required) - The MySQL username to connect with.

  • password (required) - The MySQL password to connect with.

  • address (optional) - The address of the MySQL host. Defaults to "127.0.0.1:3306.

  • database (optional) - The name of the database to use. Defaults to "vault".

  • table (optional) - The name of the table to use. Defaults to "vault".

  • tls_ca_file (optional) - The path to the CA certificate to connect using TLS

Backend Reference: PostgreSQL (Community-Supported)

The PostgreSQL backend has the following options:

  • connection_url (required) - The connection string used to connect to PostgreSQL.

    Examples:

    • postgres://username:password@localhost:5432/database?sslmode=disabled

    • postgres://username:password@localhost:5432/database?sslmode=verify-full

    A list of all supported parameters can be found in the pq library documentation.

  • table (optional) - The name of the table to write vault data to. Defaults to "vault_kv_store".

Add the following table and index to a new or existing PostgreSQL database:

CREATE TABLE vault_kv_store (
  parent_path TEXT COLLATE "C" NOT NULL,
  path        TEXT COLLATE "C",
  key         TEXT COLLATE "C",
  value       BYTEA,
  CONSTRAINT pkey PRIMARY KEY (path, key)
);

CREATE INDEX parent_path_idx ON vault_kv_store (parent_path);

If you're using a version of PostgreSQL prior to 9.5, create the following function:

CREATE FUNCTION vault_kv_put(_parent_path TEXT, _path TEXT, _key TEXT, _value BYTEA) RETURNS VOID AS
$$
BEGIN
    LOOP
        -- first try to update the key
        UPDATE vault_kv_store
          SET (parent_path, path, key, value) = (_parent_path, _path, _key, _value)
          WHERE _path = path AND key = _key;
        IF found THEN
            RETURN;
        END IF;
        -- not there, so try to insert the key
        -- if someone else inserts the same key concurrently,
        -- we could get a unique-key failure
        BEGIN
            INSERT INTO vault_kv_store (parent_path, path, key, value)
              VALUES (_parent_path, _path, _key, _value);
            RETURN;
        EXCEPTION WHEN unique_violation THEN
            -- Do nothing, and loop to try the UPDATE again.
        END;
    END LOOP;
END;
$$
LANGUAGE plpgsql;

More info can be found in the PostgreSQL documentation:

Backend Reference: Inmem

The in-memory backend has no configuration options.

Backend Reference: File

The file backend has the following options:

  • path (required) - The path on disk to a directory where the data will be stored.