Skip to content

SQLSTATE 40001 (serialization_failure) in an RPC function causes infinite retries

Last edited: 9/9/2026

A custom raise exception using SQLSTATE 40001 (serialization_failure) in your PL/pgSQL function tells PostgREST the failure is transient, so PostgREST retries the transaction. That floods your logs with duplicates and splits one API request into several independent Postgres transactions.

This bug is present in PostgREST 14 and is fixed in PostgREST 16. Follow the Supabase changelog to get notified when PostgREST 16 is released.

How to resolve#

To resolve this, update your Postgres function to use standard exception handling or SQLSTATE codes that map correctly to HTTP status codes.

1. Modify the Raise Statement Replace custom errcode assignments with a standard exception or a PostgREST-compliant code:

  • Standard Fix: Change raise exception using errcode = '40001', message = '...'; to raise exception 'YOUR_ERROR_MESSAGE';. This defaults to SQLSTATE P0001, which does not trigger the retry loop.
  • HTTP Mapping: Use the PT prefix to map to specific HTTP status codes. For example, to return an HTTP 409, use: raise sqlstate 'PT409' using message = 'YOUR_ERROR_MESSAGE';

2. Terminate Hanging Backends Existing looping processes must be stopped manually — fixing the function does not stop transactions already in flight. The repeated error entries in your Postgres logs (Logs Explorer) include a process_id field identifying the exact backend raising the error, e.g.:

"process_id": 183165,
"sql_state_code": "40001",
"user_name": "authenticator"

You can also find candidate backends in pg_stat_activity by filtering on authenticator, the role PostgREST connects as:

select pid, state, query_start, query
from pg_stat_activity
where usename = 'authenticator'
order by query_start;

Match the pid against the process_id from your logs to confirm you have the right backend, then terminate it from the SQL editor:

SELECT pg_terminate_backend(pid);

Alternatively, you can restart the project from the dashboard to clear all hanging backends at once.