Timeout seems to not work on long running queries

[mkay1375]

Expected behavior

If executing a query takes more time than request_timeout, it should be terminated with a "Timeout Exception".

Actual behavior

The execution is continued until the Trino server responds.

Steps To Reproduce

Create a connection:

from trino.dbapi import connect

connection = connect(
    host="some-trino-host",
    port="443",
    http_scheme="https",
    request_timeout=1.0
)

Execute a long running SQL which takes more than 1 second:

cur = connection.cursor()
cur.execute("SOME LONG RUNNING SQL")

Log output

No response

Operating System

macOS 15.2 (dev), Debian GNU/Linux 12 (bookworm) (prod)

Trino Python client version

0.330.0

Trino Server version

466

Python version

3.12.3 (dev), 3.12.8 (prod)

Are you willing to submit PR?

• Yes I am willing to submit a PR!

[hashhar]

the request_timeout there is the HTTP request timeout passed to Requests library. i.e. no SINGLE http request should take longer than that.

A single Trino query ends up with multiple HTTP requests. Once you submit query the client keeps polling the server periodically so as long as the server replies within 1 second the timeout will never be exceeded.

If you want to limit how long the query should run it's controllable via the session property query_max_execution_time (see https://trino.io/docs/current/admin/properties-query-management.html#query-max-execution-time). So you can set the session property on the connection/cursor to fail all queries that run longer than 1s for example.