Linux sucks

Wed Feb 03, 2010 10:13 am   

Greegor <greegor47_at_gmail.com>
wibbled on Wednesday 03 February 2010 05:57


When I flew Boeing 777's to Oz from the UK, I don't recall the in flight
film system being based (obviously) on linux - it booted straight to a
splash screen, then to the menus (though of course linux could be made to do
that).

Anyway, the crew couldn't start up every screen at once, they had to do it
in batches (quite slow to cover all seats) which implies that either the
fileserver or the LAN for these things is spec'd only to cover normal
running load and has little excess capacity for high peak loadings, for
either supplying the end device boot image or for the flurry of activity
when the thing wants its initial film list or whatever.

Given that the amount of data likely to be wanted in common should be quite
a small subset of the fileserver capacity and could easily be addressed by
more CPU and RAM (for filesystem cache) at the server end, I wonder if it's
a network limitation. My gut feeling is gigabit ethernet could trivially
handle the 220 odd seats on a 777, so perhaps they skimped on that?

If they were really clever, they'd broadcast/multicast the boot image and
initial start up files then they wouldn't have any issues with load either
on the network or the fileserver.

Didn't have any problems on the last leg (De Havilland Dash 8 Turboprop), in
flight entertainment was a newspaper and they parallelalise very efficiently
;-)

--
Tim Watts

Managers, politicians and environmentalists: Nature's carbon buffer.

Wed Feb 03, 2010 10:23 am   

G > So after reading this whole thread so far, it looks
G > like the error messages are the result of
G > the unit being deprived of a server/drive.
G >
G > That's probably not an OS problem.
G >
G > The application programmers likely
G > could have converted the somewhat
G > cryptic file error messages into a
G > more human suggestion that the
G > server is not available.
G >
G > That the crew even BOTHERS to try to
G > get it to work implies that it sometimes
G > does work, meaning it's a reliability
G > issue.
G >
G > Is it the application software people's
G > fault if there are problems accessing
G > a networked drive?
G >
G > Perhaps a network might help for the
G > output to the multiple screens on the plane
G > (working apparently) but why would they
G > want the head end to be dependent on
G > a server that seems to be so unreliable?
G >
G > If this had been a Microsoft OS computer
G > there is no guarantee that the error
G > messages would be much less cryptic.
G >
G > Regardless of the OS, Why is the
G > builder/airline tolerating such crap?
G >
G > What is the topology and how would this
G > work if it was working correctly?
G >
G > Would every monitor on the plane be
G > playing the same movie at the same time
G > or would it allow for more choices and timing?
G >
G > If done in Linux there would likely not be
G > an OS license/charge for every workstation
G > like there would if it was MS based, right?

Ban > No, they made it work after 20min. IMHO the
Ban > pilots probably had to switch on a supply for
Ban > the Video server or something that is shut
Ban > down on ground.

I suppose it's a nuisance for the flight crew
but then I assume they WANT the ability to
power off the server on take off. Perhaps the
power to the server and the power to the
workstations should both be on the same switch?


Ban > The system is quite
Ban > sophisticated, you can choose between lots
Ban > of movies or games and when there are
Ban > announcements it stops and switches over,
Ban > you can stop the film to eat and continue.

That actually sounds really slick!

JosephKK > And how many of those have been
JosephKK > posted as compared to similar conditions
JosephKK > M$ systems failing?  10,000:1?

Good point.

JosephKK > Try stupid operators suck instead.

Airline flight crew shutting down server but
leaving workstations powered?

If shutting down server is advisable
during take off, then why leave the
workstations powered up?

Is it that a parked hard disk takes the
bumps better than when the head is flying?

How would you ruggedize all of these LCD
monitors to withstand all of that vibration
and every bumpy or hard landing?

They can't get away with stock equipment
for this kind of thing, can they?

Sorry, this accidentally crept back into
the engineering realm! LOL

Wed Feb 03, 2010 10:03 pm   

On Tue, 02 Feb 2010 21:57:30 -0800, Greegor wrote:


I dunno; Windows BSOD messages are pretty cryptic, and this is
on a similar level (it's failing early in the boot process).

It's anyone's guess as to where to pin the blame. Any non-trivial boot
sequence (i.e. anything using initrd) is written by the distribution
vendor. Did the system vendor use a stock distribution or build their own?
Either way is a possibility.


Lowest bidder?


My guess: thin client with network boot. Client contains BOOTP/DHCP
and TFTP clients and nothing else. Client gets an IP address, loads the
kernel via TFTP, loads initrd (initial RAM disc, = temporary root
filesystem containing only the tools required to complete the boot
process) likewise, then mounts the real root filesystem via NFS, SMB/CIFS
or NBD (network block device).

When it works, it's great. You don't need any local drives, the only code
on the client is 8K of 8086 asm (for a PC) in a boot ROM which plugs into
the NIC, everything else is on the server. Makes administration easy, but
if the network goes down, nothing works at all.


Depends upon the capacity of the network and the servers. It makes
no difference to the client if the video stream is unique or broadcast.


Depends. The kernel, the system tools, the media player (VLC? Mplayer?),
etc don't have per-seat licence costs, and you could easily build such a
system without using any other components. Not sure what the deal would be
with MPEG (etc) licensing; may depend upon the patent laws in the country
of registration of the airliner.

Thu Feb 04, 2010 5:42 am   

On Wed, 3 Feb 2010 00:23:19 -0800 (PST), Greegor <greegor47_at_gmail.com> wrote:


Possible, especially if they are diskless.

Very much so; however, a running OS is usually writing
log files all the time. It takes hibernate to park the
heads.

Probably don't need to. Look at all the handheld stuff
with LCD screens, phones, games, media players, ebooks, etc.,

Maybe. Some careful engineering would be needed, and apparently
is not always included.

Thu Feb 04, 2010 9:06 am   

JosephKK <quiettechblue_at_yahoo.com>
wibbled on Thursday 04 February 2010 04:42


Although, with *nix, you are very much in control of that. I've had linux
boot off read only NFS, mount all of the bits it needs to write to onto
TMPFS (Ram) filesystems. Logs could be redirected to flash if they are that
important. Temporary files stay in RAM. cf "Live CDs" of various distros.

It would be very easy to have the server running and have a "takeoff" mode
where it quiesces and spins down the disks until told it is allowed to
awaken.

Though whether there is any point as boot times are a minute or two is
another question ;->

If I were designing the media streamer for an aeroplane, I'd have removeable
Flash for the OS and (possibly removable) hard disks for the films.

Life is a lot easier if the machine reliably boots and removable Flash makes
OS upgrades trivial. I don't know what procedure they currently use to add
new films, but the option to slot in a new disk's worth, and reload the old
disk on the ground I suspect has some merits.


--
Tim Watts

Managers, politicians and environmentalists: Nature's carbon buffer.

Thu Feb 04, 2010 9:08 am   

On 3/02/2010 5:44 AM, Nico Coesel wrote:

Airbuses have been very safe. Just because there was a crash of one a
while back doesn't make them unsafe.

Boeings have crashed as well.

Sylvia.

Thu Feb 04, 2010 12:59 pm   

On Thu, 04 Feb 2010 19:08:20 +1100, Sylvia Else
<sylvia_at_not.at.this.address> wrote:



You really do not understand the "Ford / Chevy" thing do you?

Thu Feb 04, 2010 9:07 pm   

Has anybody worn out any flash memory yet?

Thu Feb 04, 2010 9:48 pm   

On Feb 1, 8:02 pm, Vladimir Vassilevsky <nos...@nowhere.com> wrote:


Well it could be worse... from the cockpit: ERROR: ENGINES NOT
FOUND. SYSTEM HALTED

Michael

Fri Feb 05, 2010 1:30 am   

On Feb 2, 6:30 am, "TheM" <DontNeedS...@test.com> wrote:

Just saw that--could hardly tell it from "The Da Vinci Code." Really--
I kept wondering if we had the wrong DVD.

Neither's "Much Ado About Nothing," so to speak, but of the two "The
Da Vinci Code" gets the nod. (Ban'll like it too.)

--
Cheers,
James Arthur

Fri Feb 05, 2010 3:03 am   

On Thu, 4 Feb 2010 11:07:59 -0800 (PST), Greegor <greegor47_at_gmail.com>
wrote:



Nope. No SSD failures yet either.

Fri Feb 05, 2010 3:04 am   

On Thu, 4 Feb 2010 11:48:31 -0800 (PST), Michael <mrdarrett_at_gmail.com>
wrote:



Could be worse than that... could say "Mass Halted".


Ever watch Bugs Bunny fly a plane equipped with a Gremlin?

Fri Feb 05, 2010 7:28 am   

On Thu, 04 Feb 2010 19:08:20 +1100, Sylvia Else <sylvia_at_not.at.this.address> wrote:


By track record, the early Airbus 300s and 310s had an excessive landing
crash rates due to design vs training problems with the digital flight
controls which were very modal. This has since been resolved.

Fri Feb 05, 2010 7:36 am   

On Thu, 4 Feb 2010 11:07:59 -0800 (PST), Greegor <greegor47_at_gmail.com> wrote:


Not personally, but i have heard reports that i trust. Several year old
heavily used flash was the most common.

Fri Feb 05, 2010 8:04 am   

Mycelium wrote:


<http://www.heise.de/ct/schlagseite/2003/1/gross.jpg>

regards, Gerhard