16
$\begingroup$

On a recent flight the captain made an announcement at the gate that they need to reboot the computer as somethings aren't working, and rebooting usually solves the problem.

After posting this on twitter I got a replying saying that most planes have 3 computers and rebooting during flight is common.

What are the reasons for reboot? For me it sounds like a software/computer that isn't tested thoroughly instead.

$\endgroup$
  • 3
    $\begingroup$ writing bug-free software is hard, airplanes should not be designed so a single computer malfunction can bring it down, $\endgroup$ – ratchet freak Mar 5 '14 at 14:01
  • 9
    $\begingroup$ I wonder why the captain would feel the need to mention this to the passengers at all. $\endgroup$ – Steve V. Mar 5 '14 at 14:56
  • $\begingroup$ could he have been referring to the inertial reference system? that would be three systems, but it's not a proper computer in that sense. the mode on those can change be changed during flight i think. $\endgroup$ – Thunderstrike Mar 5 '14 at 15:37
  • 7
    $\begingroup$ "writing bug-free software is <strike>hard</strike> impossible". $\endgroup$ – abelenky Mar 5 '14 at 16:00
  • 1
    $\begingroup$ As a CS guy who had lectures from people at least somewhat involved in this field: The essential systems are generally run separately on several machines and you use some arbitration algorithm to decide what to do (this takes care of hw problems, there'll also be different software running to hopefully "take care" of sw bugs). The systems are heavily certified and generally ancient. No idea whether those machines are actually restarted during the flight if they start giving differing results. Seems like a reasonable debugging step on the ground though. $\endgroup$ – Voo Mar 6 '14 at 3:07
19
$\begingroup$

Today's modern airplanes have many computers, and sometimes things get out of sync and don't communicate properly. During the preflight checks, we check each system according to the recommendations of the manufacturer, and occasionally one of the checks will not pass correctly. Depending on the system, when a check fails, rebooting the computer may be an option to see if it fixes it (similar to rebooting your home computer). Other times, the issue is the communication protocol between two (or more) different computers. Often times in this case, the entire airplane must be powered down in order to reset them all at the same time. Since these tests aren't performed in the air, the problem usually doesn't crop up there.

In the vast majority of cases it isn't a safety critical item that we are dealing with (where the software testing is more thorough), but rather a monitoring system. The safety critical items (like flight control software, etc.) will have multiple redundant systems and often have different versions of software on at least one of the computers "just in case".

Generally the flight crew won't tell the passengers when they need to "reboot" a computer unless we have to power down the entire airplane. That's very noticeable and will cause some people to worry so we like to give you a heads up before all of the lights go out and the air conditioning system shuts down. That of course is up to the discretion of the flight crew though.

$\endgroup$
  • 1
    $\begingroup$ Fun fact, the airplanes are standard unix hosts, and usually badly outdated ones. Source $\endgroup$ – Kimvais Mar 6 '14 at 8:03
  • 15
    $\begingroup$ @Kimvais That is completely untrue. Most modern avionics components are based on embedded realtime operating systems (RTOS), some of which are based on various flavors of UNIX at a very rudimentary level. They are not "standard UNIX hosts." You can't SSH or telnet to them, get a bash shell, check your email. Your referenced source is not a primary source and I suspect its author was only exposed to avionics from a very limited perspective, and also perhaps misinterpreted or misconstrued some facts. $\endgroup$ – TypeIA Jul 8 '14 at 21:31
  • 2
    $\begingroup$ SSH, telnet, bash,and email are not proof of a UNIX system. Those are all optional IO applications that do not have to be present. That's like saying a phone is not an iPhone because the user chooses not to install and use iTunes. The proof of a UNIX system is whether it is running a UNIX kernel, and in many cases only the mfg would know. $\endgroup$ – jwzumwalt Jan 18 '18 at 19:55
11
$\begingroup$

This was probably the captain trying to be funny and use relatable terms for what was going on. In the EMB-145 occasionally the ACARS would stop working and you could re-initialize that during flight. The actual computer systems, however, could be reset but this is something you'd only do on the ground as it either involved powering down the aircraft or getting MX out to open some panels and pull breakers / flip switches.

The few things that could be reset in the air by pulling and resetting a breaker were things like FMS, ACARS and the autopilot, which we would do when directed by the QRH or MX and perhaps this is what the captain was referring to.

$\endgroup$
  • 3
    $\begingroup$ +1 and I agree completely that this was almost certainly a joke on the pilot's part. "Computers" in airplanes are not like desktop PCs with Start buttons and Reboot options. They are embedded systems in self-contained boxes buried in avionics bays in the guts of the plane. They do occasionally reset themselves due to power surges, signal transients, cosmic ray strikes (single-event upsets); but this happens entirely without pilot intervention. It is not at all a standard troubleshooting or diagnostic technique that a pilot would be employing at the gate. $\endgroup$ – TypeIA Jul 8 '14 at 21:25
5
$\begingroup$

Airliners are more complex, so I suppose it's possible. On light aircraft, I've never heard of anyone having to reboot the computer (glass panel systems). Sometimes we have to reboot our radios, and I've had to reboot FADEC (engine control) computers before though.

$\endgroup$
  • 2
    $\begingroup$ In my light aircraft, I have an Aspen Evolution glass system. I'm not sure I've never had to reboot it in flight (I might have once), but I've had a few situations where upon landing, the AHRS in it stopped working and a reboot fixed it. Had I been doing a touch-and-go rather than a full stop landing in these instances, I would have had to reboot it in flight. $\endgroup$ – mah Mar 5 '14 at 18:31
  • 3
    $\begingroup$ I've had to reboot a GNS430 in flight once. (And once was enough to convince me that having a "dumb" VHF NAV/COM in a plane is kind of essential - The 430 takes forever to reinitialize when you have to talk to someone!) $\endgroup$ – voretaq7 Mar 5 '14 at 19:02
2
$\begingroup$

Is rebooting the computer normal before/during flights?

It is wrong to write about "the computer" because a modern airliner probably contains hundreds or thousands of microprocessors and microcontrollers.

For example, the seat-back information and entertainment systems almost certainly each contain a microprocessor running some embedded operating system. Each of the items of avionics equipment will contain several microprocessors, probably running real-time operating systems that are certified for use in aircraft.

Critical avionics systems have the ability to reset themselves if they detect problems - for example they will have hardware-based watchdog timers that will restart a processor or system that isn't showing signs of running it's software normally.

Unlike a typical desktop computer, I'm pretty certain that critical systems on board an airliner are designed to handle multiple failures and keep working, in a degraded or fall-back mode if necessary.

most planes have 3 computers

As I wrote above, large commercial airliners will likely have hundreds or thousands of microprocessors and microcontrollers.

The "3 computer" idea comes from the use of redundancy in each critical system. Each subsystem may have two, three or more independently created systems with different hardware and with software written by different teams, each doing the same job. An arbitration system compares the outputs and if one of three disagrees, the other two get to determine what happens.

conclusion

No, it's not normal part of the pre-flight checklist for the pilot to walk round to the rear of the aircraft, flip a big red switch off and on again and wait for a start-up jingle.

If the in-flight movie system isn't working, I'd expect someone will try turning it off and on again.

If there is a problem with one of the many independent systems in the cockpit, I'd guess it's plausible, in some cases, the pilot or engineer's checklists for the problem include some kind of reset operation.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.