Multiple 503 errors when retrieving chart information

Summary

When loading the staging dashboard we see many 503 errors returned by prometheus

Screen_Shot_2020-07-13_at_10.40.33_AM

https://staging.gitlab.com/gitlab-org/monitor/monitor-sandbox/-/environments/180571/metrics

Steps to reproduce

  1. Visit the dashboard at https://staging.gitlab.com/gitlab-org/monitor/monitor-sandbox/-/environments/180571/metrics

  2. Set the dashboard refresh rate to 10s.

  3. Check the console error logs.

What is the current bug behavior?

After a few refreshes, multiple requests start failing with a 503 error.

The errors contain the following message:

{status: "error", message: "No prometheus server found"}

What is the expected correct behavior?

No errors are shown.

Possible fixes

Edited by Miguel Rincon