Wrong Alert's source URL

[Abuelodelanada]

Bug Description

Despite of the fact Alertmanager and Prometheus are behind an ingress, the source of the alert use k8s-based URL

To Reproduce

1. Deploy cos-lite bundle
2. In Alertmanager UI check for the Source of the alert.

Environment

• juju 3.1.6
• cos-lite bundle

Relevant log output

-

Additional context

No response

[mcarvalhor]

Affects us too.

[cbartz]

This affects us too, and also the ExternalURL (used in backlinks for the alert title in slack notifications) is set to the k8s-based URL. I wonder if changing https://github.com/canonical/alertmanager-k8s-operator/blob/main/src/charm.py#L168 to use _external_url could at least fix the ExternalURL issue.

[mcarvalhor]

Also, this may be obvious but out of interest: this wrong URL escalates to PagerDuty/monitoring services.
So on call engineers need to manually copy the link and fix the URL before pasting to the browser.

What I mean is: indeed an annoying bug, and during outages it can contribute to resolution delays.

[marcusboden]

We're also affected by this. Having to fix the URLs each time is time consuming in emergencies.

[lucabello]

Increasing the priority on this one :)