GitLab

don't maintain a persistent database connection: reconnect each time, and call initialise_database() before forking to catch errors when they will be visible on the console

pass family and proto to time_query(), rather than using the global variables, so we can eventually monitor multiple dimensions in a single process

include a node ID in database so data can be aggregated from multiple locations. pre-declare global variables and make them uppercase. sending unknown parameters causes termination

deprecate the stats file, stop accepting command line arguments except --config, --debug and --help. allow cached address records to expire and be refreshed

include protocol and family in the database spec so multiple instances can write to the same database. update documentation