The customer had done some failure over test for their MX960-VC chassis, the outcome as below:
Offline/Online an FPC and measure (~5 second outage which we are investigating)
Graceful RE switchover (subsecond outage)
Graceful VC switchover (subsecond outage)
VMHost reboot (subsecond outage)
They would like to check whether this is an expected behavior.
Based on the research, it is expected that there will be few seconds outage if the customer force an FPC offline either by the CLI or by removing it physically. There is an internal watchdog daemon to check the status of the line card, the system need wait for the timeout.