Actual drop time and duration are [1447085423, 1447085481), and 58s.
Though IFO seemed to keep a full lock according to LSC_LOCK_STATE_N in this segment, some guardian nodes reported "not OK". Checking the details of guardian logs around that time, then connection errors of EPICS records provided by K1IX1 and K1IY1 were reported. (It may be a good news to that it hasn't been an unapproved silence injection.)
Because it's a strange to make error on K1IX1 and K1IY1 at the same time, I doubted the network switch instead of each front-end. Checking logs of the network switch at 2F of the corner station then uptime of this switch shows 11h 19m at 12:30 JST. It's consistent with the error timing. So the network disconnection of K1IX1 and K1IY1 due to a hang-up and a reboot of the network switch seemed to make a drop from the science mode.
