Skip to content
Database

pg_net: Async Networking

pg_net enables Postgres to make asynchronous HTTP/HTTPS requests in SQL. It differs from the http extension in that it is asynchronous by default. This makes it useful in blocking functions (like triggers).

It eliminates the need for servers to continuously poll for database changes and instead allows the database to proactively notify external resources about significant events.

Enable the extension#

  1. Go to the Database page in the Dashboard.
  2. Click on Extensions in the sidebar.
  3. Search for "pg_net" and enable the extension.

http_get#

Creates an HTTP GET request returning the request's ID. HTTP requests are not started until the transaction is committed.

Signature #

net.http_get(
    -- url for the request
    url text,
    -- key/value pairs to be url encoded and appended to the `url`
    params jsonb default '{}'::jsonb,
    -- key/values to be included in request headers
    headers jsonb default '{}'::jsonb,
    -- the maximum number of milliseconds the request may take before being canceled
    timeout_milliseconds int default 2000
)
    -- request_id reference
    returns bigint
    strict
    volatile
    parallel safe
    language plpgsql

Usage #

select
    net.http_get('https://news.ycombinator.com')
    as request_id;
request_id
----------
         1
(1 row)

http_post#

Creates an HTTP POST request with a JSON body, returning the request's ID. HTTP requests are not started until the transaction is committed.

The body's character set encoding matches the database's server_encoding setting.

Signature #

net.http_post(
    -- url for the request
    url text,
    -- body of the POST request
    body jsonb default '{}'::jsonb,
    -- key/value pairs to be url encoded and appended to the `url`
    params jsonb default '{}'::jsonb,
    -- key/values to be included in request headers
    headers jsonb default '{"Content-Type": "application/json"}'::jsonb,
    -- the maximum number of milliseconds the request may take before being canceled
    timeout_milliseconds int default 2000
)
    -- request_id reference
    returns bigint
    volatile
    parallel safe
    language plpgsql

Usage #

select
    net.http_post(
        url:='https://httpbin.org/post',
        body:='{"hello": "world"}'::jsonb
    ) as request_id;
request_id
----------
         1
(1 row)

http_delete#

Creates an HTTP DELETE request, returning the request's ID. HTTP requests are not started until the transaction is committed.

Signature #

net.http_delete(
    -- url for the request
    url text,
    -- key/value pairs to be url encoded and appended to the `url`
    params jsonb default '{}'::jsonb,
    -- key/values to be included in request headers
    headers jsonb default '{}'::jsonb,
    -- the maximum number of milliseconds the request may take before being canceled
    timeout_milliseconds int default 2000
)
    -- request_id reference
    returns bigint
    strict
    volatile
    parallel safe
    language plpgsql
    security definer

Usage #

select
    net.http_delete(
        'https://dummy.restapiexample.com/api/v1/delete/2'
    ) as request_id;
----------
         1
(1 row)

Analyzing responses#

Waiting requests are stored in the net.http_request_queue table. Upon execution, they are deleted.

CREATE UNLOGGED TABLE
    net.http_request_queue (
        id bigint NOT NULL DEFAULT nextval('net.http_request_queue_id_seq'::regclass),
        method text NOT NULL,
        url text NOT NULL,
        headers jsonb NOT NULL,
        body bytea NULL,
        timeout_milliseconds integer NOT NULL
    )

Once a response is returned, by default, it is stored for 6 hours in the net._http_response table.

CREATE UNLOGGED TABLE
    net._http_response (
        id bigint NULL,
        status_code integer NULL,
        content_type text NULL,
        headers jsonb NULL,
        content text NULL,
        timed_out boolean NULL,
        error_msg text NULL,
        created timestamp with time zone NOT NULL DEFAULT now()
    )

The responses can be observed with the following query:

select * from net._http_response;

The data can also be observed in the net schema with the Supabase Dashboard's SQL Editor

Debugging requests#

Inspecting request data#

The Postman Echo API returns a response with the same body and content as the request. It can be used to inspect the data being sent.

Sending a post request to the echo API

select
    net.http_post(
        url := 'https://postman-echo.com/post',
        body := '{"key1": "value", "key2": 5}'::jsonb
    ) as request_id;

Inspecting the echo API response content to ensure it contains the right body

select
    "content"
from net._http_response
where id = <request_id>
-- returns information about the request
-- including the body sent: {"key": "value", "key": 5}

Alternatively, by wrapping a request in a database function, sent row data can be logged or returned for inspection and debugging.

create or replace function debugging_example (row_id int)
returns jsonb as $$
declare
    -- Store payload data
    row_data_var jsonb;
begin
    -- Retrieve row data and convert to JSON
    select to_jsonb("<example_table>".*) into row_data_var
    from "<example_table>"
    where "<example_table>".id = row_id;
    -- Initiate HTTP POST request to URL
    perform
        net.http_post(
            url := 'https://postman-echo.com/post',
            -- Use row data as payload
            body := row_data_var
        ) as request_id;
    -- Optionally Log row data or other data for inspection in Supabase Dashboard's Postgres Logs
    raise log 'Logging an entire row as JSON (%)', row_data_var;
    -- return row data to inspect
    return row_data_var;
-- Handle exceptions here if needed
exception
    when others then
        raise exception 'An error occurred: %', SQLERRM;
end;
$$ language plpgsql;
-- calling function
select debugging_example(<row_id>);

Inspecting failed requests#

Finds all failed requests

select
  *
from net._http_response
where "status_code" >= 400 or "error_msg" is not null
order by "created" desc;

Configuration#

The extension is configured to reliably execute up to 200 requests per second. The response messages are stored for only 6 hours to prevent needless buildup. The default behavior can be modified by rewriting config variables.

Get current settings#

select
  "name",
  "setting"
from pg_settings
where "name" like 'pg_net%';

Alter settings#

You must change the pg_net settings at the system level.

Once the privilege is assigned, apply the setting at the system level and restart the background worker:

alter system set pg_net.ttl to '24 hours';
select net.worker_restart();

Examples#

Invoke a Supabase Edge Function#

Make a POST request to a Supabase Edge Function with auth header and JSON body payload:

select
    net.http_post(
        url:='https://project-ref.supabase.co/functions/v1/function-name',
        headers:='{"Content-Type": "application/json", "apikey": "<SUPABASE_PUBLISHABLE_KEY>"}'::jsonb,
        body:='{"name": "pg_net"}'::jsonb
    ) as request_id;

Call an endpoint every minute with pg_cron#

The pg_cron extension enables Postgres to become its own cron server. With it you can schedule regular calls with up to a minute precision to endpoints.

select cron.schedule(
	'cron-job-name',
	'* * * * *', -- Executes every minute (cron syntax)
	$$
	    -- SQL query
	    select "net"."http_post"(
            -- URL of Edge function
            url:='https://project-ref.supabase.co/functions/v1/function-name',
            headers:='{"apikey": "<SUPABASE_PUBLISHABLE_KEY>"}'::jsonb,
            body:='{"name": "pg_net"}'::jsonb
	    ) as "request_id";
	$$
);

Execute pg_net in a trigger#

Make a call to an external endpoint when a trigger event occurs.

-- function called by trigger
create or replace function <function_name>()
    returns trigger
    language plpgSQL
as $$
begin
    -- calls pg_net function net.http_post
    -- sends request to postman API
    perform "net"."http_post"(
      'https://postman-echo.com/post'::text,
      jsonb_build_object(
        'old_row', to_jsonb(old.*),
        'new_row', to_jsonb(new.*)
      ),
      headers:='{"Content-Type": "application/json"}'::jsonb
    ) as request_id;
    return new;
END $$;
-- trigger for table update
create trigger <trigger_name>
    after update on <table_name>
    for each row
    execute function <function_name>();

Send multiple table rows in one request#

with "selected_table_rows" as (
    select
        -- Converts all the rows into a JSONB array
        jsonb_agg(to_jsonb(<table_name>.*)) as JSON_payload
    from <table_name>
    -- good practice to LIMIT the max amount of rows
)
select
    net.http_post(
        url := 'https://postman-echo.com/post'::text,
        body := JSON_payload
    ) AS request_id
FROM "selected_table_rows";

More examples can be seen on the Extension's GitHub page

Permissions#

By default, the net schema grants USAGE to PUBLIC. As a result, anon and authenticated inherit direct object-level access (for example, SELECT) on net.http_request_queue, net._http_response, and their associated sequences.

This doesn't expose request data to unauthenticated or client-side users, for two reasons:

  • net isn't exposed through the Data API, so anon/publishable keys can't access or modify its objects through the API.
  • anon and authenticated are NOLOGIN roles, so they can't establish a direct database connection.

Limitations#

  • To improve speed and performance, the requests and responses are stored in unlogged tables, which are not preserved during a crash or unclean shutdown.
  • By default, response data is saved for only 6 hours
  • Can only make POST requests with JSON data. No other data formats are supported
  • Intended to handle at most 200 requests per second. Increasing the rate can introduce instability
  • Does not have support for PATCH/PUT requests
  • Can only work with one database at a time. It defaults to the postgres database.

Resources#