Notes that probably need to end up in release notes:
Peak updates happening close to real time are considered a sign of a healthy wallet.
If the wallet peak stays unhelthy for the configured WALLET keep-alive threshold, a keep-alive alert is triggered. By default this is 5 minutes.
If your node is not running a WALLET, remove it from your config.yamlmonitored_services section. If you don't yet have that section, have a peek at config-example.yaml. By default the following services are assumed to be running: FULL_NODE, FARMER, WALLET, HARVESTER.
The default and example value of peak drift is set at 2 minutes. This is based on observation under current load conditions. I'm guessing a dust storm or the like may cause this to spike and trigger an unhealthy wallet keep-alive if the wallet can't keep up. Time will tell if that default needs to be adjusted to be more permissive. Also this was observed on a fairly beefy node, so running on a RasPi will likely need the value adjusted if the wallet constantly lags by more than 2 minutes.
Notes that probably need to end up in release notes:
config.yaml
monitored_services
section. If you don't yet have that section, have a peek atconfig-example.yaml
. By default the following services are assumed to be running: FULL_NODE, FARMER, WALLET, HARVESTER.