Hello Percona Support Team,
We are currently using PMM3 in our production environment, and I’ve run into an issue after attempting to route Grafana alerts through an internal proxy.
- Initially, we were unable to send Slack notifications via the built-in
grafana-alertmanagerdue to internal network restrictions (outbound traffic blocked by proxy). - As a workaround, we deployed an external Alertmanager Docker container and registered it as a data source in PMM Grafana.
- The custom alert rules and messaging pipeline are now functioning correctly, and Slack notifications are being delivered as expected.
- Additionally, all of our custom dashboards are working fine — metrics appear correctly because they explicitly use the
Metricsdata source. - However, the built-in dashboards that come with PMM3 do not show any metrics. They appear blank.
- Upon investigation, I found that these built-in dashboards are using
-- Grafana --as their data source, which appears to be linked to synthetic/fake data. - I’m not sure why the data source has changed for the built-in dashboards — this was not the case before. I tried:
- Reverting to the previous configuration by disabling the external Alertmanager,
- Re-selecting
Grafanaunder “Contact Points → Choose Alertmanager”, - Even modifying container-mounted dashboard JSON files.None of these attempts resolved the issue.
Is there a supported way to restore or rebind the built-in dashboards so that they point to the correct Metrics data source again (instead of -- Grafana --)?
Any advice or documented method to recover this behavior would be greatly appreciated.
Thank you for your support.
[This is a built-in dashboards]
