Use-cases:

• snmp-ups with large devices (daisy-chains etc.) can take literally minutes to walk the MIB and initialize the first state after which it would regularly do smaller updates

  • more so if the device is not accessible

• dummy-ups can take a while to parse the data file, interrupting for TIMER lines etc. and then would stop reading it (with dummy-once mode since dummy-ups loops over definition file regardless of "seq" or "dev" naming #1385) - hence initialization complete

Currently in automated testing we have to rely on timing or casual guess to decide if we can interrogate the driver and have it some expected state. To an extent, the "Error: driver not connected" handles the bare initialization but does not tell us if the driver is ready to give stable reports. In a way, that is the moment where a "dump mode" with -d 1 option would exit the driver.

A similar (but probably different) useful dstate option could reflect if the driver is currently parsing external data (began/finished the regular update), also during non-initializing loops.