OpenML task API returns 503 for task 31, causing ope... | openml/OpenML

Nobody has claimed this yet.

Title: OpenML task API returns 503 for task 31, causing openml-python / HPOBench task loading to fail

Hi OpenML team,

I am trying to use OpenML task 31 through openml-python and HPOBench, but the API endpoint currently returns 503 Service Temporarily Unavailable.

The failing endpoint is:

https://www.openml.org/api/v1/xml/task/31

I tested it directly with curl:

curl -L -i https://www.openml.org/api/v1/xml/task/31 | head -n 30

The response is:

HTTP/1.1 503 Service Temporarily Unavailable
Date: Tue, 02 Jun 2026 08:22:33 GMT
Server: Apache/2.4.41 (Ubuntu)
Content-Type: text/html
Content-Length: 190
Strict-Transport-Security: max-age=31536000; includeSubDomains
Connection: close

<html>
<head><title>503 Service Temporarily Unavailable</title></head>
<body>
<center><h1>503 Service Temporarily Unavailable</h1></center>
<hr><center>nginx</center>
</body>
</html>

This also causes openml-python to fail when calling:

import openml
task = openml.tasks.get_task(31, download_data=False)

In my HPOBench workflow, the error appears when initializing:

from hpobench.benchmarks.ml import LRBenchmark

task_id = 31
seed = 0
bench = LRBenchmark(task_id=task_id, rng=seed)

The traceback shows that the local task cache is missing, so openml-python tries to fetch the task description from the API. However, the API returns an HTML 503 page instead of the expected XML response, which then leads to an XML parsing failure:

OpenMLCacheException: Task file for tid 31 not cached

...

openml.exceptions.OpenMLServerError: Unexpected server error when calling https://www.openml.org/api/v1/xml/task/31.
Status code: 503

<html>
<head><title>503 Service Temporarily Unavailable</title></head>
<body>
<center><h1>503 Service Temporarily Unavailable</h1></center>
<hr><center>nginx</center>
</body>
</html>

Environment:

OS: macOS
Python: 3.9
openml-python: installed through the current Python environment
HPOBench: local clone / third_party HPOBench

Could you please check whether the task API endpoint is currently down or whether task 31 is temporarily unavailable?

Thanks!

Start by reproducing the reported request with curl against https://www.openml.org/api/v1/xml/task/31, then verify the same failure through openml.tasks.get_task(31, download_data=False). Check the task API service responsible for this endpoint. Done means task 31 returns the expected XML response and the openml-python and HPOBench examples can load it without a 503 or parsing error.

Written by the indexing model from the issue text.

OpenML task API returns 503 for task 31, causing openml-python / HPOBench task loading to fail

Description

Contributor guide

First steps

Research direction

Assessment