Self-hosted grafana instances on prefix path cannot be connected >1.14.4

[MiddleMan5]

What went wrong?

What happened:
When upgrading from 1.13.3 to 1.14.4 the plugin installation fails. After digging into the code it appears that the grafanaUrl that is configured from the grafana ui settings is sanitized incorrectly before it is stored in the database here:

oncall/engine/apps/grafana_plugin/serializers/sync_data.py

Line 78 in b507941

return value.rstrip("/")

Plugin installation fails with "Plugin Not Connected"

and following error logs:

2025-05-07 19:39:10 source=engine:app google_trace_id=none logger=apps.grafana_plugin.helpers.client Error connecting to api instance 404 Client Error: Not Found for url: http://localhost:3000/api/org

Note grafana url is now missing /grafana path prefix

This appears to be caused by the python urljoin call here which ignores path prefixes without a trailing slash

See test below:

from urllib.parse import urljoin

print(urljoin("http://localhost:3000/grafana", "api/org"))
print(urljoin("http://localhost:3000/grafana/", "api/org"))

Output

http://localhost:3000/api/org
http://localhost:3000/grafana/api/org

How do we reproduce it?

1. Deploy grafana on a /grafana path prefix
2. Install and configure oncall plugin with localhost:3000/grafana/ note trailing slash
3. Plugin installation fails with "Plugin Not Connected"

Grafana OnCall Version

1.14.4

Product Area

API

Grafana OnCall Platform?

Kubernetes

User's Browser?

No response

Anything else to add?

No response

[MiddleMan5]

Confirmed workaround is to monkey patch /etc/app/apps/grafana_plugin/serializers/sync_data.py file in the docker image and add a forward slash at the end of the grafana_url in the validate_grafana_url function.

Oncall is now able to successfully connect