Describe the bug

I have a Github actions job to check for broken links in the API docs for the Aerospike Python client, and currently there are false positive timeouts that often occur. To mitigate this, I tried 1) increasing the timeout period to 120 seconds, as well as 2) increasing the number of retries to 4. Neither approach seems to fix the problem.

Workflow results increasing retries to 4, while keeping timeout to 30 seconds: https://github.com/aerospike/aerospike-client-python/actions/runs/22962838493/job/66658057652?pr=989

When setting retries to 4, linkcheck takes 55 seconds to run, so it makes me wonder if it is retrying properly. I cannot tell from reading the logs with -vv set.

How to Reproduce

git clone https://github.com/aerospike/aerospike-client-python
cd aerospike-client-python/doc
# python -m venv .venv
# . .venv/bin/activate
pip install -r requirements.txt
sphinx-build -b linkcheck -vv . links

Environment Information

Platform:              linux; (Linux-6.8.0-1044-azure-x86_64-with-glibc2.35)
Python version:        3.12.12 (main, Oct 10 2025, 00:59:52) [GCC 11.4.0])
Python implementation: CPython
Sphinx version:        9.1.0
Docutils version:      0.22.4
Jinja2 version:        3.1.6
Pygments version:      2.19.2

Sphinx extensions

[
    "sphinx.ext.todo",
    "sphinx.ext.autodoc",
    "sphinx.ext.intersphinx",
    "sphinx.ext.napoleon",
    "sphinxcontrib.spelling"
]

Additional context

If I manually check the links by clicking on them, the browser will show them as okay.