Adding new metric to service performance data breaks graphing

We’re using nagios-plugin-cloudwatch to graph data from Cloudwatch. It supports getting multiple metrics (statistics) from AWS and converts them into Nagios-compatible performance data. On a basic level it works well.

Two days ago I added a new metric (BurstBalance) to an existing service check. There were two problems:

  • The graph boxes for the new metric (BurstBalance) never appeared
  • All of the old metrics stopped graphing (I now have two-day gap in the old metrics)

After creating an identical, new LibreNMS service graphing started working properly, including the new metric and all the old ones.

Is this a known issue? Is there a way to “refresh” a Service so that it does not choke if performance data changes, e.g. when a new metric is added.

Our LibreNMS instance is based on commit 0040f38d1ed1f7caac60326f72ad61555d69b4f5 (June 30th 2020).