Restarting Application creates a new metric each time

[jseiser]

We have a python application, started by systemd. When we restart the app for any reason, we end up with a new metric being created ( assume because the PID value changes)

when I query active_tunnels ill get something like this back, where each PID the application ever had, is its own metric. This makes the metric rather useless since I can not easily say how many active tunnels do we have

active_tunnels{instance="Foundation_xxxx_gateway_srv01", instanceip="Foundation_xxxx_gateway_srv01:10.55.1.15", ip="10.55.1.15", job="gw_exporter", layer="Gateway", pid="1244"} 0
active_tunnels{instance="Foundation_xxxx_gateway_srv01", instanceip="Foundation_xxxx_gateway_srv01:10.55.1.15", ip="10.55.1.15", job="gw_exporter", layer="Gateway", pid="15961"} 2
active_tunnels{instance="Foundation_xxxx_gateway_srv01", instanceip="Foundation_xxxx_gateway_srv01:10.55.1.15", ip="10.55.1.15", job="gw_exporter", layer="Gateway", pid="17418"} 2
active_tunnels{instance="Foundation_xxxx_gateway_srv01", instanceip="Foundation_xxxx_gateway_srv01:10.55.1.15", ip="10.55.1.15", job="gw_exporter", layer="Gateway", pid="2730"} 2
active_tunnels{instance="Foundation_xxxx_gateway_srv01", instanceip="Foundation_xxxx_gateway_srv01:10.55.1.15", ip="10.55.1.15", job="gw_exporter", layer="Gateway", pid="2929"} 2

There is another prom client running ont his same server, which uses flask/gunicorn, so its setup for the multiproc stuff. My guess is this is somehow interferring?

[jseiser]

OK, there is a gunicorn flask app on this same server, and it is what is exposing the metrics, so I think we need to actually configure this for multiproc support.