Authentication

PostgREST is designed to keep the database at the center of API security. All authorization happens in the database . It is PostgREST’s job to authenticate requests – i.e. verify that a client is who they say they are – and then let the database authorize client actions.

Overview of role system

There are three types of roles used by PostgREST, the authenticator, anonymous and user roles. The database administrator creates these roles and configures PostgREST to use them.

The authenticator role is used for connecting to the database and should be configured to have very limited access. It is a chameleon whose job is to “become” other users to service authenticated HTTP requests.

CREATE ROLE authenticator LOGIN NOINHERIT NOCREATEDB NOCREATEROLE NOSUPERUSER;
CREATE ROLE anonymous NOLOGIN;
CREATE ROLE webuser NOLOGIN;

Note

The names “authenticator” and “anon” names are configurable and not sacred, we simply choose them for clarity. See db-uri and db-anon-role.

User Impersonation

The picture below shows how the server handles authentication. If auth succeeds, it switches into the user role specified by the request, otherwise it switches into the anonymous role (if it’s set in db-anon-role).

This role switching mechanism is called user impersonation. In PostgreSQL it’s done with the SET ROLE statement.

Note

The impersonated roles will have their settings applied. See Impersonated Role Settings.

JWT Authentication

We use JSON Web Tokens to authenticate API requests, this allows us to be stateless and not require database lookups for verification. As you’ll recall a JWT contains a list of cryptographically signed claims. All claims are allowed but PostgREST cares specifically about a claim called role (configurable with JWT Role Extraction).

{
  "role": "user123"
}

When a request contains a valid JWT with a role claim PostgREST will switch to the database role with that name for the duration of the HTTP request.

SET LOCAL ROLE user123;

Note that the database administrator must allow the authenticator role to switch into this user by previously executing

GRANT user123 TO authenticator;
-- similarly for the anonymous role
-- GRANT anonymous TO authenticator;

If the client included no JWT (or one without a role claim) then PostgREST switches into the anonymous role. The database administrator must set the anonymous role permissions correctly to prevent anonymous users from seeing or changing things they shouldn’t.

Bearer Authentication

To make an authenticated request the client must include an Authorization HTTP header with the value Bearer <jwt>. For instance:

curl "http://localhost:3000/foo" \
  -H "Authorization: Bearer eyJhbGciOiJIUzI1NiIsInR5cCI6IkpXVCJ9.eyJyb2xlIjoiamRvZSIsImV4cCI6MTQ3NTUxNjI1MH0.GYDZV3yM0gqvuEtJmfpplLBXSGYnke_Pvnl0tbKAjB4"

The Bearer header value can be used with or without capitalization(bearer).

JWT Generation

You can create a valid JWT either from inside your database (see SQL User Management) or via an external service (see External Authentication).

JWT Signature Verification

PostgREST supports both symmetric and asymmetric keys for verifying the signature of the token.

Symmetric Keys

In the case of symmetric cryptography the signer and verifier share the same secret passphrase, which can be configured with jwt-secret. If it is set to a simple string then PostgREST interprets it as an HMAC-SHA256 passphrase.

jwt-secret = "reallyreallyreallyreallyverysafe"

Asymmetric Keys

In asymmetric cryptography the signer uses the private key and the verifier the public key.

As described in the Configuration section, PostgREST accepts a jwt-secret config file parameter. However you can also specify a literal JSON Web Key (JWK) or set. For example, you can use an RSA-256 public key encoded as a JWK:

{
  "alg":"RS256",
  "e":"AQAB",
  "key_ops":["verify"],
  "kty":"RSA",
  "n":"9zKNYTaYGfGm1tBMpRT6FxOYrM720GhXdettc02uyakYSEHU2IJz90G_MLlEl4-WWWYoS_QKFupw3s7aPYlaAjamG22rAnvWu-rRkP5sSSkKvud_IgKL4iE6Y2WJx2Bkl1XUFkdZ8wlEUR6O1ft3TS4uA-qKifSZ43CahzAJyUezOH9shI--tirC028lNg767ldEki3WnVr3zokSujC9YJ_9XXjw2hFBfmJUrNb0-wldvxQbFU8RPXip-GQ_JPTrCTZhrzGFeWPvhA6Rqmc3b1PhM9jY7Dur1sjYWYVyXlFNCK3c-6feo5WlRfe1aCWmwZQh6O18eTmLeT4nWYkDzQ"
}

Note

This could also be a JSON Web Key Set (JWKS) if it was contained within an array assigned to a keys member, e.g. { keys: [jwk1, jwk2] }.

Just pass it in as a single line string, escaping the quotes:

jwt-secret = "{ \"alg\":\"RS256\", … }"

To generate such a public/private key pair use a utility like latchset/jose.

jose jwk gen -i '{"alg": "RS256"}' -o rsa.jwk
jose jwk pub -i rsa.jwk -o rsa.jwk.pub

# now rsa.jwk.pub contains the desired JSON object

You can specify the literal value as we saw earlier, or reference a filename to load the JWK from a file:

jwt-secret = "@rsa.jwk.pub"

`kid` verification

PostgREST has built-in verification of the key ID parameter, useful when working with a JSON Web Key Set. It goes as follows:

If the JWT contains a kid parameter, then PostgREST will look for the JSON Web Key in the jwt-secret.
- If no key has a matching kid (or if they don’t have one defined), the token will be rejected with a 401 Unauthorized error.
- If a key matches the kid value then it will validate the token against that key accordingly.
If the JWT doesn’t have a kid, PostgREST will try each key in the jwt-secret one by one until it finds one that works.

JWT Claims Validation

Time-Based claims validation

The time-based JWT claims specified in RFC 7519 are validated:

exp Expiration Time
iat Issued At
nbf Not Before

We allow a 30-second clock skew when validating the above claims. In other words, we give an extra 30 seconds before the JWT is rejected if there is a slight discrepancy in the timestamps.

`aud` validation

PostgREST has built-in validation of the JWT audience claim. It works this way:

If jwt-aud is not set (the default), PostgREST identifies with all audiences and allows the JWT for any aud claim.
If jwt-aud is set to a specific audience, PostgREST will check if this audience is present in the aud claim:
- If the aud value is a JSON string, it will match it to the jwt-aud.
- If the aud value is a JSON array of strings, it will search every element for a match.
- If the match fails or if the aud value is not a string or array of strings, then the token will be rejected with a 401 Unauthorized error.
- If the aud key is not present or if its value is null or [], PostgREST will interpret this token as allowed for all audiences and will complete the request.

JWT Cache

JWT signature validation (specially Asymmetric Keys such as RSA) is slow, we can cache JWT validation results to avoid this performance overhead.

The JWT cache is bounded and uses the SIEVE algorithm for efficient eviction. The cache is enabled by default and can be configured with jwt-cache-max-entries.

It’s recommended to leave the JWT cache enabled as our load tests indicate ~20% more throughput for simple GET requests when using it. This while reducing CPU utilization in exchange for a bit more memory.

JWT Cache Metrics are available.

Note

If the jwt-secret is changed and the config is reloaded, the JWT cache will reset.
JWTs that pass JWT Signature Verification are cached, regardless if they pass JWT Claims Validation. We do this to ensure responses stays fast under common failure cases (such as expired JWTs).
You can use the Server-Timing Header to see the peformance benefit of JWT caching.

JWT Role Extraction

A JSPath DSL that specifies the location of the role key in the JWT claims. It’s configured by jwt-role-claim-key. This can be used to consume a JWT provided by a third party service like Auth0, Okta, Microsoft Entra or Keycloak.

The DSL follows the JSONPath expression grammar with extended string comparison operators. Supported operators are:

== selects the first array element that exactly matches the right operand
!= selects the first array element that does not match the right operand
^== selects the first array element that starts with the right operand
==^ selects the first array element that ends with the right operand
*== selects the first array element that contains the right operand

Usage examples:

# {"postgrest":{"roles": ["other", "author"]}}
# the DSL accepts characters that are alphanumerical or one of "_$@" as keys
jwt-role-claim-key = ".postgrest.roles[1]"

# {"https://www.example.com/role": { "key": "author" }}
# non-alphanumerical characters can go inside quotes(escaped in the config value)
jwt-role-claim-key = ".\"https://www.example.com/role\".key"

# {"postgrest":{"roles": ["other", "author"]}}
# `@` represents the current element in the array
# all the these match the string "author"
jwt-role-claim-key = ".postgrest.roles[?(@ == \"author\")]"
jwt-role-claim-key = ".postgrest.roles[?(@ != \"other\")]"
jwt-role-claim-key = ".postgrest.roles[?(@ ^== \"aut\")]"
jwt-role-claim-key = ".postgrest.roles[?(@ ==^ \"hor\")]"
jwt-role-claim-key = ".postgrest.roles[?(@ *== \"utho\")]"

Note

The string comparison operators are implemented as a custom extension to the JSPath and does not strictly follow the RFC 9535.

JWT Security

There are at least three types of common critiques against using JWT: 1) against the standard itself, 2) against using libraries with known security vulnerabilities, and 3) against using JWT for web sessions. We’ll briefly explain each critique, how PostgREST deals with it, and give recommendations for appropriate user action.

The critique against the JWT standard is voiced in detail elsewhere on the web. The most relevant part for PostgREST is the so-called alg=none issue. Some servers implementing JWT allow clients to choose the algorithm used to sign the JWT. In this case, an attacker could set the algorithm to none, remove the need for any signature at all and gain unauthorized access. The current implementation of PostgREST, however, does not allow clients to set the signature algorithm in the HTTP request, making this attack irrelevant. The critique against the standard is that it requires the implementation of the alg=none at all.

Another type of critique focuses on the misuse of JWT for maintaining web sessions. The basic recommendation is to stop using JWT for sessions because most, if not all, solutions to the problems that arise when you do, do not work. The linked articles discuss the problems in depth but the essence of the problem is that JWT is not designed to be secure and stateful units for client-side storage and therefore not suited to session management.

PostgREST uses JWT mainly for authentication and authorization purposes and encourages users to do the same. For web sessions, using cookies over HTTPS is good enough and well catered for by standard web frameworks.

Custom Validation

PostgREST does not enforce any extra constraints besides JWT validation. An example of an extra constraint would be to immediately revoke access for a certain user. Using db-pre-request you can specify a function to call immediately after User Impersonation and before the main query itself runs.

db-pre-request = "public.check_user"

In the function you can run arbitrary code to check the request and raise an exception(see RAISE errors with HTTP Status Codes) to block it if desired. Here you can take advantage of Request Headers, Cookies and JWT claims for doing custom logic based on the web user info.

CREATE OR REPLACE FUNCTION check_user() RETURNS void AS $$
DECLARE
  email text := current_setting('request.jwt.claims', true)::json->>'email';
BEGIN
  IF email = 'evil.user@malicious.com' THEN
    RAISE EXCEPTION 'No, you are evil'
      USING HINT = 'Stop being so evil and maybe you can log in';
  END IF;
END
$$ LANGUAGE plpgsql;

Authentication

Overview of role system

User Impersonation

JWT Authentication

Bearer Authentication

JWT Generation

JWT Signature Verification

Symmetric Keys

Asymmetric Keys

kid verification

JWT Claims Validation

Time-Based claims validation

aud validation

JWT Cache

JWT Role Extraction

JWT Security

Custom Validation

`kid` verification

`aud` validation