Is your feature request related to a problem? Please describe.
The metrics dashboard shows inaccurate values when the node is re-started for the number of connected peers, and the usage of the mempool. This is due to cummulative nature of the prometheus metrics.

Describe the solution you'd like
A simple explicit resetting of these metrics to 0 when the node starts/restarts, should fix this.

Describe alternatives you've considered
Not doing anything, and keep things as they are, hoping that someone will remember that this is how the metrics work.

Additional context
Having metrics that show unreliable numbers has caused interpretation problems in the past.