open-vault/website/source/docs/auth/app-id.html.md

112 lines
3.9 KiB
Markdown
Raw Normal View History

2015-04-18 20:45:50 +00:00
---
layout: "docs"
page_title: "Auth Backend: App ID"
sidebar_current: "docs-auth-appid"
description: |-
The App ID auth backend is a mechanism for machines to authenticate with Vault.
---
# Auth Backend: App ID
Name: `app-id`
The App ID auth backend is a mechanism for machines to authenticate with
Vault. It works by requiring two hard-to-guess unique pieces of information:
a unique app ID, and a unique user ID.
The goal of this credential provider is to allow elastic users
(dynamic machines, containers, etc.) to authenticate with Vault without
having to store passwords outside of Vault. It is a single method of
solving the chicken-and-egg problem of setting up Vault access on a machine.
With this provider, nobody except the machine itself has access to both
pieces of information necessary to authenticate. For example:
configuration management will have the app IDs, but the machine itself
will detect its user ID based on some unique machine property such as a
MAC address (or a hash of it with some salt).
An example, real world process for using this provider:
1. Create unique app IDs (UUIDs work well) and map them to policies.
(Path: map/app-id/<app-id>)
2. Store the app IDs within configuration management systems.
3. An out-of-band process run by security operators map unique user IDs
to these app IDs. Example: when an instance is launched, a cloud-init
system tells security operators a unique ID for this machine. This
process can be scripted, but the key is that it is out-of-band and
out of reach of configuration management.
(Path: map/user-id/<user-id>)
2015-04-18 20:45:50 +00:00
4. A new server is provisioned. Configuration management configures the
app ID, the server itself detects its user ID. With both of these
pieces of information, Vault can be accessed according to the policy
set by the app ID.
More details on this process follow:
The app ID is a unique ID that maps to a set of policies. This ID is
generated by an operator and configured into the backend. The ID itself
is usually a UUID, but any hard-to-guess unique value can be used.
After creating app IDs, an operator authorizes a fixed set of user IDs
2015-04-29 05:04:58 +00:00
with each app ID. When a valid {app ID, user ID} tuple is given to the
"login" path, then the user is authenticated with the configured app
ID policies.
2015-04-18 20:45:50 +00:00
The user ID can be any value (just like the app ID), however it is
generally a value unique to a machine, such as a MAC address or instance ID,
or a value hashed from these unique values.
## Authentication
#### Via the CLI
App ID authentication is not allowed via the CLI.
#### Via the API
2015-05-15 18:32:48 +00:00
The endpoint for the App ID login is `/login`. The client is expected
to provide the `app_id` and `user_id` parameters as part of the request.
2015-04-18 20:45:50 +00:00
## Configuration
First you must enable the App ID auth backend:
```
$ vault auth-enable app-id
Successfully enabled 'app-id' at 'app-id'!
```
Now when you run `vault auth -methods`, the App ID backend is available:
```
Path Type Description
app-id/ app-id
token/ token token based credentials
```
2015-04-18 20:45:50 +00:00
To use the App ID auth backend, an operator must configure it with
the set of App IDs, user IDs, and the mapping between them. An
example is shown below, use `vault help` for more details.
```
$ vault write auth/app-id/map/app-id/foo value=root display_name=foo
...
$ vault write auth/app-id/map/user-id/bar value=foo cidr_block=10.0.0.0/16
2015-04-18 20:45:50 +00:00
...
```
The above creates an App ID "foo" that associates with the policy "root".
The `display_name` sets the display name for audit logs and secrets.
Next, we configure the user ID "bar" and say that the user ID bar
can be paired with "foo" but only in client is in the "10.0.0.0/16" CIDR block.
The `cidr_block` configuration is optional.
2015-04-18 20:45:50 +00:00
This means that if a client authenticates and provide both "foo" and "bar",
then the app ID will authenticate that client with the policy "root".
In practice, both the user and app ID are likely hard-to-guess UUID-like values.