With the hardware absent, a daemon sits in its TCP connect before registering any keywords, so it is active under systemd but answers nothing for the whole window. Measured 2m13s for atcpress on hispec-new (15:25:27 start, 15:27:40 "started peer=atcpress").
During that window libby list <group>.<daemon>.% just times out, which looks identical to a broken daemon. A connect timeout on the driver sockets, or registering keywords before attempting hardware, would make this diagnosable.
With the hardware absent, a daemon sits in its TCP connect before registering any keywords, so it is
activeunder systemd but answers nothing for the whole window. Measured 2m13s for atcpress on hispec-new (15:25:27 start, 15:27:40 "started peer=atcpress").During that window
libby list <group>.<daemon>.%just times out, which looks identical to a broken daemon. A connect timeout on the driver sockets, or registering keywords before attempting hardware, would make this diagnosable.